Unattended conveyor system
By integrating centralized control system, AI video surveillance system and network broadcasting system, using machine vision technology and YOLOv5 model, real-time detection and alarm of belt conveyors are achieved, and the problem of insufficient detection accuracy and accuracy in the existing technology is solved. Improves the efficiency and safety of belt detection.
Patent Information
- Application Number
- CN202311184384.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-09-14
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2043-09-14
AI Technical Summary
When the existing belt conveyor detection system detects belt tear, the accuracy and accuracy are insufficient, and the slight tear cannot be detected in time, resulting in serious damage to the belt and serious safety hazards.
Adopt centralized control system, AI video surveillance system and network broadcasting system to detect belt conditions in real time through industrial cameras and lasers, and use machine vision technology and YOLOv5 convolutional neural network model to identify belt tear and deviation, realizing automatic alarm and shutdown protection .
It improves the accuracy and accuracy of belt detection, can detect in real time during belt operation, reduces manual inspection costs, improves detection efficiency and safety, and prevents further damage to the belt.
Smart Images

Figure CN117184814B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to an automated production system, in particular to an unattended conveyor system. Background Art
[0002] At present, the main solution in the field of industrial automation control is PLC, which provides very reliable control applications for various automation control equipment. The main reason is that it can provide safe, reliable and relatively complete solutions for automation control applications, which is suitable for the current needs of enterprises for automation. The belt conveyor centralized control system, dispatch telephone system, industrial television system and network broadcasting system together constitute a complete operation, dispatch and monitoring network, realizing the telemetry, telecommunication and remote control of the entire transportation system.
[0003] Fieldbus is the most widely used industrial control technology in today's enterprises, but the biggest problem with fieldbus is that there are too many standards. What enterprises need is to establish a unified and open network, and the monitoring and surveillance equipment of different enterprises can be connected to this network for transmission. Obviously, if fieldbus is used, the device-level network used to connect sensors can still be competent, but as a backbone network for data transmission, it cannot meet our needs in terms of transmission rate, transmission distance and link redundancy, and it is bound to make the transmission network transition from an incompatible multi-system structure to another incompatible multi-system structure, which obviously does not meet the needs of building modern factories and realizing comprehensive automation. With the rapid development of the Internet, Ethernet has become the de facto industrial standard, and TCP / IP is simple and practical, and is accepted by the majority of users.
[0004] Ethernet has unparalleled advantages over other networks in many aspects such as technology, speed and price. With the improvement of Ethernet performance and the continuous introduction of technologies to solve Ethernet real-time problems, the application of Ethernet in industrial sites will be an inevitable choice in the field of industrial control. At present, industrial Ethernet has been put into practical application on the ground. Enterprises use industrial Ethernet as the backbone transmission network for belt conveyor control and fieldbus as the network for connecting field equipment such as sensors. This should be the research focus and construction direction of factory transmission networks in the near future. It can be foreseen that industrial Ethernet will be the inevitable choice for the construction of enterprise broadband communication networks in the future. This belt conveyor unattended system is a new generation of unattended system that applies industrial Ethernet to belt conveyors. In traditional production sites, there are no examples of applying existing or newly created technologies alone or in combination to conveyors.
[0005] Belt conveyors are important tools and equipment for transporting coal in ports, mines, power plants and other enterprises. Longitudinal tearing accidents of conveyor belts caused by impurities occur from time to time. If not discovered in time, the entire belt will be torn and damaged, causing huge economic losses. In order to solve this technical problem, domestic and foreign experts have tried to use many methods for anti-tear detection, such as traditional impact detection method, abnormal roller force detection, ultrasonic method, varistor method, embedding method, etc. However, these methods currently have the problem of high maintenance costs and cumbersome maintenance steps, which bothers the staff.
[0006] The Chinese patent with application number 202310054344.X provides a machine vision belt tear detection and protection device, which is provided with a light source sensing plate at the bottom of the conveyor belt, and a laser emitting device is fixedly provided on both sides of the top of the conveyor belt. The laser emitting device emits laser to illuminate the conveyor belt. When the conveyor belt is torn, the laser will pass through the torn gap of the conveyor belt and illuminate the light source sensing plate. When the light source sensing plate senses the laser light, it will immediately transmit the signal to the laser device, and the laser device will transmit the signal to the host inside the main box. The host will immediately alarm to notify the staff and shut down the belt conveyor, thereby preventing further tearing and breaking of the conveyor belt.
[0007] There are still disadvantages in actual use: the detection precision and accuracy are not enough, and it is impossible to detect when the belt has a slight tear. When the belt is detected to be torn, the damage to the belt is already very serious, posing a serious safety hazard. Summary of the invention
[0008] In view of the deficiencies in the above-mentioned prior art, the present invention provides an unmanned conveyor system and method, particularly a control system and method for preventing and preventing conveyor belt deviation and tearing with high detection precision and accuracy.
[0009] The technical solution adopted by the present invention is:
[0010] An unmanned conveyor system, comprising: a centralized control system, an AI video monitoring system and a network broadcasting system;
[0011] The centralized control system includes: a host computer, a network cabinet, a PLC master station, a distributed IO substation, and a sensor; the host computer realizes remote centralized control; the sensor realizes signal acquisition, specifically including:
[0012] The network fault self-diagnosis module can automatically detect and issue an alarm when the network has transmission problems such as disconnection or interference.
[0013] PLC fault self-diagnosis module: When the PLC scanner and adapter fail, the fault is determined through the network communication status and an alarm is issued; when the PLC I / O module fails, the fault and fault content are detected through the status bit of the I / O module, and the system will issue an alarm;
[0014] Sensor and signal line fault diagnosis module: when the analog sensor or signal line is disconnected, the PLC determines the fault through the measured value and issues an alarm;
[0015] The I / O module uses a detachable terminal block. When a fault occurs, the control room will issue an audible and visual alarm.
[0016] The webcast system includes:
[0017] The central control room is equipped with a main switch and a broadcast control host;
[0018] The system has all-area calling and broadcasting, partition calling and broadcasting;
[0019] The network broadcasting system can freely set the audio content, time and date of the broadcast, and has automatic partition broadcasting and compatibility with multiple external media. It can realize arbitrary selection of playback, point-to-point voice intercom, point-to-face voice intercom, hard disk memory is convenient for the storage of audio information, and adopts an open sound source software platform to add / record audio information as needed by users.
[0020] AI video surveillance system includes:
[0021] Conveyor belt deviation and tear control system; used to determine whether the conveyor belt is in the correct working position during operation and whether there is tearing that affects material transportation;
[0022] Conveyor belt deviation and tear control system; obtain the belt detection model through training, and then compare the collected belt pictures with the different feature image data in the belt detection model through real-time detection to obtain the actual belt condition. Use machine vision real-time detection to replace manual inspection, and real-time detection can be performed during the operation of the belt;
[0023] Conveyor belt loading detection system; used to determine whether the belt conveyor is loaded normally or empty;
[0024] Foreign matter identification system: used to detect the particle size of materials on the conveyor belt in the inspection path, determine whether there are foreign matters in the materials by the particle size, and accurately locate them;
[0025] Personnel identification system: used to identify personnel in the inspection area; identify and expel non-staff members from the designated area; analyze, judge, remind and alarm on illegal operations and illegal contact with equipment by on-site personnel;
[0026] Instrument recognition system: automatically recognizes the readings, indicator light status, switch status, and mechanical position of various instruments through image recognition algorithms.
[0027] Preferably, the conveyor belt deviation and tearing control system is used to determine whether the conveyor belt is in the correct working position during operation and whether tearing that affects material transportation occurs. The control method includes the following steps:
[0028] Step s400, starting the device;
[0029] Step s401, turn on the industrial camera to obtain the belt image in real time;
[0030] Step s402, the camera starts the continuous capture mode;
[0031] Step s403, converting the grayscale image into a binary image according to a set threshold;
[0032] Step s404, determine whether the belt is torn; if the belt is torn, jump to step s405; if the belt is not torn, jump to step s410.
[0033] Step s405, determining whether the belt is torn at the second level; if the belt is torn at the second level, jump to step s406; if the belt is not torn at the second level, jump to step s408;
[0034] Step s406, sending a shutdown command to the controller, sounding an alarm, and recording a secondary tear in the fault log;
[0035] Step s407, the belt conveyor stops running, the device is turned off, and the process goes to step s416;
[0036] Step s408, sounding an alarm and recording a first-level tear in the fault record;
[0037] Step s409, execute the tearing control system sub-process; jump to step s116;
[0038] Step s410, determining whether the belt is deviated; if the belt is not deviated, jump to step s401; if the belt is deviated, jump to step s411;
[0039] Step s411, determining whether the belt has a secondary deviation; if the belt has a secondary deviation, jump to step s412; if the belt has not reached a secondary deviation, jump to step s414;
[0040] Step s412, sending a stop command to the controller, sounding an alarm, and recording the secondary deviation in the fault record;
[0041] Step s413, the belt conveyor stops running and the device is turned off; jump to step s416;
[0042] Step s414, sounding an alarm and recording a first-level deviation in the fault record;
[0043] Step s415, execute the deviation control system sub-process; jump to step s416;
[0044] Step s416, end.
[0045] The conveyor belt longitudinal tear detection device provides a truly reliable, practical and scientific detection method for belt conveyor users to detect longitudinal tear of the belt. It can conduct all-weather real-time online detection of the bottom condition of the upper belt when the belt conveyor is operating normally. Once the belt is longitudinally torn, an alarm will be issued immediately (within 20 meters of the longitudinal tear) and the belt operation will be stopped to prevent the accident from further expansion. The detection and identification of longitudinal tear are all intelligent, with automatic identification and automatic alarm. The product has the characteristics of high sensitivity, high reliability, simple structure, easy installation and maintenance, and suitable for harsh environments.
[0046] The optical triangulation principle and echo analysis principle are used for imaging, and a precision sensor is used to measure the width and depth of the belt tear crack without contact. The laser and camera are respectively installed on the left and right lower sides of the upper belt. The laser emits a laser to the bottom of the belt, which is collected by the camera after being reflected on the belt surface, and the center of the laser stripe is automatically extracted. The height and depth parameters of the belt section are analyzed according to the triangulation distance measurement method, and output to the control computer in real time. After computer software analysis and data restoration, a 3D image of the belt bottom is obtained. It can intuitively determine whether the belt is torn.
[0047] The belt surface condition is automatically judged and analyzed in real time, and the belt surface damage such as debonding, scratches, and tears are identified and judged respectively, and different alarm treatments are given. New and old damages can be identified, and the identification of new and old damages can be set to avoid repeated secondary alarms and shutdowns.
[0048] Real-time monitoring: The belt running status is displayed in real time on the computer screen, and the belt surface image can be viewed smoothly.
[0049] Centralized control interlocking: can implement simultaneous sound and light alarms on site and in the control room according to the damage situation; when the damage is greater than the length, width, and depth values set by the software, it will control the emergency stop protection shutdown and alarm at the same time.
[0050] Early warning: Early warning for damages where the tear length or width is less than the set value, damages where the covering adhesive falls off, damages where the edge tape is damaged, etc.
[0051] Playback storage: After the tearing machine stops, the local enlarged image of the tearing position can be viewed and saved.
[0052] High stability: The equipment is required to run continuously during the operation of the belt, and the Ethernet (optical fiber) communication network is used, with high data communication reliability.
[0053] Preferably, in step s409, the tearing control system sub-process steps are as follows:
[0054] Step s500, tearing control system sub-process entry;
[0055] Step s501, PLC receives a tearing signal;
[0056] Step s502, determine whether it is a secondary tear, if it is a secondary tear, jump to step s504, otherwise jump to step s503;
[0057] Step s503, PLC only sends out an alarm signal; jump to step s510;
[0058] Step s504, PLC issues a command and the belt conveyor stops;
[0059] Step s505, waiting for a manual signal, determining whether the belt conveyor is restarted, if restarted, jump to step s506, otherwise jump to step s508;
[0060] Step s506, sending a power-on signal;
[0061] Step s507, the belt conveyor starts running, and jumps to step s510;
[0062] Step s508, no power-on signal is issued;
[0063] Step s509, the belt conveyor stops and waits for manual maintenance, and jumps to step s510;
[0064] Step s510, tearing the control system sub-process exit.
[0065] Preferably, the steps of the deviation control system sub-process in step s415 are as follows:
[0066] Step s600, entry of the deviation control system sub-process;
[0067] Step s601, PLC receives a deviation signal;
[0068] Step s602, determine whether it is a secondary deviation, if it is a secondary deviation, jump to step s605, otherwise jump to step s603;
[0069] Step s603, PLC only sends out an alarm signal;
[0070] Step s604, the automatic hydraulic deviation correction device intervenes to correct the belt position, and jumps to step s607;
[0071] Step s605, PLC issues a command and the belt conveyor stops;
[0072] Step s606, waiting for a manual signal, determining whether the belt conveyor is restarted, if restarted, jump to step s607, otherwise jump to step s609;
[0073] Step s607, sending a power-on signal;
[0074] Step s608, the belt conveyor starts running, and jumps to step s611;
[0075] Step s609, no power-on signal is issued;
[0076] Step s610, the belt conveyor stops and waits for manual maintenance, and jumps to step s611;
[0077] Step s611, exit of the deviation control system sub-process.
[0078] Preferably, the 1,000 collected image samples are manually marked to mark the tearing features and deviation features, and compared with the normal belt. The collected images are divided into a training set: validation set: test set ratio of 7:2:1 using the YOLOv5 convolutional neural network, and the model with weight YOLOv5s is used for training, 500 rounds, and batch-size 4. The training results are obtained and the model with the best training degree is selected. This model is the belt detection model.
[0079] Preferably, step s104, the steps of determining whether the belt is deviated are as follows:
[0080] The industrial camera obtains the normal working state picture of the belt, and converts the normal working state picture grayscale picture into a binary picture according to the set threshold value; sets the normal working area of the belt;
[0081] The purpose of converting the grayscale image into a binary image is to better highlight the outline of the laser auxiliary line. The grayscale image is 8 bits, and the value ranges from 0-255, 0 is pure black, and 255 is pure white. In order to maximize the display of the laser and facilitate subsequent image recognition and processing, a threshold will be set according to the on-site situation. Assuming the threshold is 180, all values above 180 are converted to 255, and all values below 180 are converted to 0. The image will become a binary image with only black and white colors. At that time, the white line will become very obvious, which is convenient for the machine to recognize later.
[0082] Compare the real-time captured image binary image with the normal working state image binary image to determine whether the belt is deviating.
[0083] Real-time snapshot images are mainly used to extract features for training. In order to expand the training set and database, as many deviation images with obvious features as possible are collected to better train the belt deviation model and meet the characteristics of strong adaptability as much as possible. In this way, it can meet as many usage scenarios as possible and no longer needs to be trained separately for each belt. The real-time snapshot images and the actual working status images are taken from the same angle by the camera. They are mainly compared through some deviation features, such as detecting the edge of the belt to see if the edge is deviated, such as detecting whether the roller frame is exposed, etc. to determine whether it is deviated.
[0084] The threshold value set here is set according to the on-site environment. The purpose is to make the line illuminated by the laser on the belt more obvious. Setting the normal operating area of the belt means mainly setting the tear detection area near the white line irradiated by the laser.
[0085] Preferably, the high-speed camera takes 30 photos in a high-speed continuous shooting.
[0086] Industrial cameras usually shoot belts in a stationary state. During the operation of the conveyor belt, the belt load is in motion. The belt speed on the market is often 0-10m / s. The purpose of the high-speed camera to capture 30 pictures per second is to capture 30 pictures per second when the conveyor belt is moving at 10m / s, so that the pictures taken will not be blurred.
[0087] Preferably, whether the belt is torn is determined by using a SIFT corner point detection method.
[0088] Preferably, it comprises: a central processing unit and an inspection robot; the inspection robot comprises: an intelligent detection platform, and an industrial camera is arranged on the intelligent detection platform;
[0089] The inspection robot is a hanging rail walking robot;
[0090] The hanging rail of the hanging rail walking robot is set above the belt conveyor.
[0091] The beneficial effects of the present invention compared with the prior art are as follows:
[0092] The unattended conveyor system of the present invention has more complete control, more reliable protection, more convenient operation, more convenient maintenance, easier expansion and networking; improves the speed of troubleshooting; has stable and reliable performance, and is flexible in expansion.
[0093] The unattended conveyor system of the present invention, the control system, dispatch telephone system, industrial television system, network broadcasting system together constitute a complete operation, dispatch, monitoring network, realize the telemetry, telecommunication, remote control of the entire transportation system. By receiving the alarm information of the LED splicing screen, it is possible to timely understand which belt conveyor and which position has a specific abnormal situation.
[0094] The unattended conveyor system of the present invention and the factory's fully digital AI video surveillance can conduct real-time monitoring, real-time recording, and real-time data storage of the operating conditions of all equipment in the factory (such as belt conveyor deviation, temperature during equipment operation, etc.) and the dynamic flow of personnel to achieve scientific safety management. The video surveillance monitoring system is an important component of the security technology prevention system. It is an advanced, comprehensive system with strong prevention capabilities. It can directly view all conditions at the work site through remote control cameras and their auxiliary equipment; the image content of the work site can be transmitted to the monitoring center in real time, so that the center dispatchers can see the on-site conditions at a glance, monitor whether the entire operation process is standardized, and avoid the occurrence of safety hazards in time.
[0095] The unattended conveyor system of the present invention utilizes an industrial video monitoring system, so that monitoring personnel in the central control room can directly monitor the on-site situation in real time, which can not only intuitively monitor and record the safety production situation at the work site, but also promptly discover the seeds of accidents and prevent them before they happen; at the same time, it can record all or part of the images and sounds of the monitored places, and can also provide relevant first-hand image data for subsequent analysis of accidents, thus providing convenient conditions and important basis for the future handling of certain events. At the same time, the television monitoring system can also be linked with other security technical prevention systems such as anti-theft alarms to make the prevention capability more powerful.
[0096] The conveyor belt deviation and tearing control method of the present invention obtains a belt detection model through training, and then compares the collected belt pictures with different feature image data in the belt detection model through real-time detection to obtain the actual belt condition. Real-time detection by machine vision replaces manual inspection, and real-time detection can be performed during the operation of the belt, which greatly reduces labor costs and improves detection efficiency and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS
[0097] Figure 1 It is a centralized control network system diagram of the unattended conveyor system of the present invention;
[0098] Figure 2 It is a schematic diagram of a network broadcasting system of an unattended conveyor system of the present invention;
[0099] Figure 3 It is a diagram of the AI video monitoring system of the unattended conveyor system of the present invention;
[0100] Figure 4 It is a control flow diagram of a conveyor belt deviation and tearing control method of an unattended conveyor system of the present invention;
[0101] Figure 5It is a schematic diagram of a tearing subroutine control flow of a conveyor belt deviation and tearing control method of an unattended conveyor system of the present invention;
[0102] Figure 6 The present invention is a schematic diagram of a control flow of a conveyor belt deviation subroutine of a conveyor belt deviation and tearing control method of an unattended conveyor system of the present invention. DETAILED DESCRIPTION
[0103] The present invention is described in detail below with reference to the accompanying drawings and embodiments:
[0104] A conveyor unattended system includes: a centralized control system, an AI video monitoring system and a network broadcasting system.
[0105] Attached Figure 1 The centralized control system includes: host computer, network cabinet, PLC master station, distributed IO substation, sensor; the host computer realizes remote centralized control; the sensor realizes signal acquisition, specifically including:
[0106] The network fault self-diagnosis module can automatically detect and issue an alarm when the network has transmission problems such as disconnection or interference.
[0107] PLC fault self-diagnosis module: When the PLC scanner and adapter fail, the fault is determined through the network communication status and an alarm is issued; when the PLC I / O module fails, the fault and fault content are detected through the status bit of the I / O module, and the system will issue an alarm;
[0108] Sensor and signal line fault diagnosis module: when the analog sensor or signal line is disconnected, the PLC determines the fault through the measured value and issues an alarm;
[0109] The I / O module uses a detachable terminal block. When a fault occurs, the control room will issue an audible and visual alarm.
[0110] The replacement can be completed within a few minutes. The system is easy to expand. When adding new equipment or later projects, nodes can be easily added, and the program can be modified online through the network. The system has a self-diagnosis function and can report fault time, location, and type information in a timely manner.
[0111] Attached Figure 2 , the network broadcasting system includes:
[0112] The central control room is equipped with main switch, ring network switch, video IP phone, dispatch host, broadcast control host, and broadcast server;
[0113] The system has all-area calling and broadcasting, partition calling and broadcasting;
[0114] The network broadcasting system can freely set the audio content, time and date of the broadcast, and has automatic partition broadcasting and compatibility with multiple external media. It can realize arbitrary selection of playback, point-to-point voice intercom, point-to-face voice intercom, hard disk memory is convenient for the storage of audio information, and adopts an open sound source software platform to add / record audio information as needed by users.
[0115] Attached Figure 3 , AI video surveillance system includes:
[0116] Conveyor belt deviation and tear control system; used to determine whether the conveyor belt is in the correct working position during operation and whether there is tearing that affects material transportation;
[0117] Conveyor belt deviation and tear control system; obtain the belt detection model through training, and then compare the collected belt pictures with the different feature image data in the belt detection model through real-time detection to obtain the actual belt condition. Use machine vision real-time detection to replace manual inspection, and real-time detection can be performed during the operation of the belt;
[0118] Conveyor belt loading detection system; used to determine whether the belt conveyor is loaded normally or empty;
[0119] Foreign object identification system: used to detect the particle size of materials on the conveyor belt in the inspection path, determine whether there are foreign objects in the materials by the particle size, and accurately locate them to avoid damage to the equipment.
[0120] Personnel identification system: used to identify personnel in the inspection area; identify and expel non-staff members from the designated area; analyze, judge, remind and alarm on illegal operations and illegal contact with equipment by on-site personnel;
[0121] Instrument recognition system: automatically recognizes the readings, indicator light status, switch status, and mechanical position of various instruments through image recognition algorithms.
[0122] Attached Figure 4 , conveyor belt deviation and tearing control system; the control method for determining whether the conveyor belt is in the correct working position during operation and whether tearing that affects material transportation occurs includes the following steps:
[0123] Step s400, starting the device;
[0124] Step s401, turn on the industrial camera to obtain the belt image in real time;
[0125] Step s402, the camera starts the continuous capture mode;
[0126] Step s403, converting the grayscale image into a binary image according to a set threshold;
[0127] Step s404, determine whether the belt is torn; if the belt is torn, jump to step s405; if the belt is not torn, jump to step s410.
[0128] Step s405, determining whether the belt is torn at the second level; if the belt is torn at the second level, jump to step s406; if the belt is not torn at the second level, jump to step s408;
[0129] Step s406, sending a shutdown command to the controller, sounding an alarm, and recording a secondary tear in the fault log;
[0130] Step s407, the belt conveyor stops running, the device is turned off, and the process goes to step s416;
[0131] Step s408, sounding an alarm and recording a first-level tear in the fault record;
[0132] Step s409, execute the tearing control system sub-process; jump to step s416;
[0133] Step s410, determining whether the belt is deviated; if the belt is not deviated, jump to step s401; if the belt is deviated, jump to step s411;
[0134] Step s411, determining whether the belt has a secondary deviation; if the belt has a secondary deviation, jump to step s412; if the belt has not reached a secondary deviation, jump to step s414;
[0135] Step s412, sending a stop command to the controller, sounding an alarm, and recording the secondary deviation in the fault record;
[0136] Step s413, the belt conveyor stops running and the device is turned off; jump to step s416;
[0137] Step s414, sounding an alarm and recording a first-level deviation in the fault record;
[0138] Step s415, execute the deviation control system sub-process; jump to step s416;
[0139] Step s416, end.
[0140] During the operation of the belt, tears usually do not occur within the specified service life. Tearing will only occur when the service life is exceeded or severe wear occurs. Tearing usually occurs first and then second. The development of first-level tear to second-level tear is a slow process. However, in actual use, when the belt is overloaded or there are a large number of sharp objects in the carried goods, the time for the first-level tear to develop to the second-level tear is very short. Therefore, when it is determined that the belt has a first-level tear, in order to ensure the safe operation of the belt, the tear control system sub-process needs to make another determination.
[0141] Preferably, in step s409, the tearing control system sub-process steps are as follows:
[0142] Step s500, tearing control system sub-process entry
[0143] Step s501, PLC receives a tearing signal;
[0144] Step s502, determine whether it is a secondary tear, if it is a secondary tear, jump to step s504, otherwise jump to step s503;
[0145] Step s503, PLC only sends out an alarm signal; jump to step s510;
[0146] Step s504, PLC issues a command and the belt conveyor stops;
[0147] Step s505, waiting for a manual signal, determining whether the belt conveyor is restarted, if restarted, jump to step s506, otherwise jump to step s508;
[0148] Step s506, sending a power-on signal;
[0149] Step s507, the belt conveyor starts running, and jumps to step s510;
[0150] Step s508, no power-on signal is issued;
[0151] Step s509, the belt conveyor stops and waits for manual maintenance, and jumps to step s510;
[0152] Step s510, tearing the control system sub-process exit.
[0153] During the operation of the belt, it usually does not deviate within the specified service life. It will only deviate after a certain period of use or severe wear. Deviation usually occurs first and then second. The development of first deviation to second is a slow process. However, when the belt is overloaded during actual use, the time it takes for the first deviation to develop to second is very short. Therefore, when it is determined that the belt has first deviation, in order to ensure the safety of the belt operation, the deviation control system sub-process needs to make another determination.
[0154] Preferably, the steps of the deviation control system sub-process in step s415 are as follows:
[0155] Step s600, entry of the deviation control system sub-process;
[0156] Step s601, PLC receives a deviation signal;
[0157] Step s602, determine whether it is a secondary deviation, if it is a secondary deviation, jump to step s605, otherwise jump to step s603;
[0158] Step s603, PLC sends out an alarm signal and the belt conveyor stops;
[0159] Step s604, the automatic hydraulic deviation correction device intervenes to correct the belt position, and jumps to step s607;
[0160] Step s605, PLC issues a command and the belt conveyor stops;
[0161] Step s606, waiting for a manual signal, determining whether the belt conveyor is restarted, if restarted, jump to step s607, otherwise jump to step s609;
[0162] Step s607, sending a power-on signal;
[0163] Step s608, the belt conveyor starts running, and jumps to step s611;
[0164] Step s609, no power-on signal is issued;
[0165] Step s610, the belt conveyor stops and waits for manual maintenance, and jumps to step s611;
[0166] Step s611, exit of the deviation control system sub-process.
[0167] Preferably, the 1,000 collected image samples are manually marked to mark the tearing features and deviation features, and compared with the normal belt. The collected images are divided into a training set: validation set: test set ratio of 7:2:1 using the YOLOv5 convolutional neural network, and the model with weight YOLOv5s is used for training, 500 rounds, and batch-size 4. The training results are obtained and the model with the best training degree is selected. This model is the belt detection model.
[0168] The industrial camera obtains the normal working state picture of the belt, and converts the normal working state picture grayscale picture into a binary picture according to the set threshold value; sets the normal working area of the belt;
[0169] The purpose of converting the grayscale image into a binary image is to better highlight the outline of the laser auxiliary line. The grayscale image is 8 bits, and the value ranges from 0-255, 0 is pure black, and 255 is pure white. In order to maximize the display of the laser and facilitate subsequent image recognition and processing, a threshold will be set according to the on-site situation. Assuming the threshold is 180, all values above 180 are converted to 255, and all values below 180 are converted to 0. The image will become a binary image with only black and white colors. At that time, the white line will become very obvious, which is convenient for the machine to recognize later.
[0170] Compare the real-time captured image binary image with the normal working state image binary image to determine whether the belt is deviating.
[0171] Real-time snapshot images are mainly used to extract features for training. In order to expand the training set and database, as many deviation images with obvious features as possible are collected to better train the belt deviation model and meet the characteristics of strong adaptability as much as possible. In this way, it can meet as many usage scenarios as possible and no longer needs to be trained separately for each belt. The real-time snapshot images and the actual working status images are taken from the same angle by the camera. They are mainly compared through some deviation features, such as detecting the edge of the belt to see if the edge is deviated, such as detecting whether the roller frame is exposed, etc. to determine whether it is deviated.
[0172] The threshold value set here is set according to the on-site environment. The purpose is to make the line illuminated by the laser on the belt more obvious. Setting the normal operating area of the belt means mainly setting the tear detection area near the white line irradiated by the laser.
[0173] Preferably, the high-speed camera takes 30 photos in a high-speed continuous shooting.
[0174] Industrial cameras usually shoot belts in a stationary state. During the operation of the conveyor belt, the belt load is in motion. The belt speed on the market is often 0-10m / s. The purpose of the high-speed camera to capture 30 pictures per second is to capture 30 pictures per second when the conveyor belt is moving at 10m / s, so that the pictures taken will not be blurred.
[0175] Preferably, whether the belt is torn is determined by using a SIFT corner point detection method.
[0176] Preferably, it comprises: a central processing unit and an inspection robot; the inspection robot comprises: an intelligent detection platform, and an industrial camera is arranged on the intelligent detection platform;
[0177] The inspection robot is a hanging rail walking robot;
[0178] The hanging rail of the hanging rail type walking robot is arranged above the belt conveyor.
[0179] The conveyor belt deviation and tearing control method of the present invention obtains a belt detection model through training, and then compares the collected belt pictures with different feature image data in the belt detection model through real-time detection to obtain the actual belt condition. Real-time detection by machine vision replaces manual inspection, and real-time detection can be performed during the operation of the belt, which greatly reduces labor costs and improves detection efficiency and accuracy.
[0180] The above is only a preferred embodiment of the present invention, and does not limit the structure of the present invention in any form. Any simple modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention are within the scope of the technical solution of the present invention.
Claims
1. An unattended conveyor system, characterized in that: Including centralized control system, AI video surveillance system and network broadcasting system; The centralized control system includes: a host computer, a network cabinet, a PLC master station, a distributed IO substation, and a sensor; the host computer realizes remote centralized control; the sensor realizes signal acquisition, specifically including: The network fault self-diagnosis module can automatically detect and issue an alarm when the network has transmission problems such as disconnection or interference. PLC fault self-diagnosis module: When the PLC scanner and adapter fail, the fault is determined through the network communication status and an alarm is issued; when the PLC I / O module fails, the fault and fault content are detected through the status bit of the I / O module, and the system will issue an alarm; Sensor and signal line fault diagnosis module: when the analog sensor or signal line is disconnected, the PLC determines the fault through the measured value and issues an alarm; The I / O module uses a detachable terminal block. When a fault occurs, the control room will issue an audible and visual alarm. The webcast system includes: The central control room is equipped with a main switch and a broadcast control host; The system has all-area calling and broadcasting, partition calling and broadcasting; The network broadcasting system can freely set the audio content, time and date of the broadcast, and has automatic partition broadcasting and compatibility with multiple external media. It can realize arbitrary selection of playback, point-to-point voice intercom, point-to-face voice intercom, hard disk memory is convenient for the storage of audio information, and adopts an open sound source software platform to add / record audio information as needed by users. AI video surveillance system includes: The conveyor belt deviation and tear control system is used to determine whether the conveyor belt is in the correct working position during operation and whether there is tearing that affects material transportation; Conveyor belt loading detection system; used to determine whether the belt conveyor is loaded normally or empty; The foreign body identification system is used to detect the particle size of the material on the conveyor belt in the inspection path, determine whether there are foreign bodies in the material by the particle size, and accurately locate them; Personnel identification system: used to identify personnel in the inspection area; identify and expel non-staff members from the designated area; analyze, judge, remind and alarm on illegal operations and illegal contact with equipment by on-site personnel; Instrument recognition system: automatically recognizes the readings, indicator light status, switch status, and mechanical position of various instruments through image recognition algorithms; The conveyor belt deviation and tearing control system is used to determine whether the conveyor belt is in the correct working position during operation and whether tearing that affects material transportation occurs, and includes the following steps: Step s400, starting the device; Step s401, turn on the industrial camera to obtain the belt image in real time; Step s402, the camera starts the continuous capture mode; Step s403, converting the grayscale image into a binary image according to a set threshold; Step s404, determine whether the belt is torn; if the belt is torn, jump to step s405; if the belt is not torn, jump to step s410. Step s405, determining whether the belt is torn at the second level; if the belt is torn at the second level, jump to step s406; if the belt is not torn at the second level, jump to step s408; Step s406, sending a shutdown command to the controller, sounding an alarm, and recording a secondary tear in the fault log; Step s407, the belt conveyor stops running, the device is turned off, and the process goes to step s416; Step s408, sounding an alarm and recording a first-level tear in the fault record; Step s409, execute the tearing control system sub-process; jump to step s116; Step s410, determining whether the belt is deviated; if the belt is not deviated, jump to step s401; if the belt is deviated, jump to step s411; Step s411, determining whether the belt has a secondary deviation; if the belt has a secondary deviation, jump to step s412; if the belt has not reached a secondary deviation, jump to step s414; Step s412, sending a stop command to the controller, sounding an alarm, and recording the secondary deviation in the fault record; Step s413, the belt conveyor stops running and the device is turned off; jump to step s416; Step s414, sounding an alarm and recording a first-level deviation in the fault record; Step s415, execute the deviation control system sub-process; jump to step s416; Step s416, end; The optical triangulation principle and echo analysis principle are used for imaging. A precision sensor is used to non-contactly measure the width and depth of the belt tear crack. The laser and camera are respectively installed on the left and right lower sides of the upper belt. The laser emits a laser to the bottom of the belt. After being reflected on the belt surface, it is collected by the camera and the center of the laser stripe is automatically extracted. The height and depth parameters of the belt section are analyzed according to the triangulation distance measurement method and output to the control computer in real time. After computer software analysis and data restoration, a 3D image of the belt bottom is obtained, which can intuitively determine whether the belt is torn. Automatically judge and analyze the belt surface condition in real time, identify and judge the belt surface damage such as degumming, scratches, and tears, and give different alarm treatments. It can identify new and old damages, and can set the identification of new and old damages without repeated secondary alarms and shutdowns; The steps of step s409, the tearing control system sub-process are as follows: Step s500, tearing control system sub-process entry; Step s501, PLC receives a tearing signal; Step s502, determine whether it is a secondary tear, if it is a secondary tear, jump to step s504, otherwise jump to step s503; Step s503, PLC only sends out an alarm signal; jump to step s510; Step s504, PLC issues a command and the belt conveyor stops; Step s505, waiting for a manual signal, determining whether the belt conveyor is restarted, if restarted, jump to step s506, otherwise jump to step s508; Step s506, sending a power-on signal; Step s507, the belt conveyor starts running, and jumps to step s510; Step s508, no power-on signal is issued; Step s509, the belt conveyor stops and waits for manual maintenance, and jumps to step s510; Step s510, tearing the control system sub-process exit; In step s415, the steps of the deviation control system sub-process are as follows: Step s600, entry of the deviation control system sub-process; Step s601, PLC receives a deviation signal; Step s602, determine whether it is a secondary deviation, if it is a secondary deviation, jump to step s605, otherwise jump to step s603; Step s603, PLC only sends out an alarm signal; Step s604, the automatic hydraulic deviation correction device intervenes to correct the belt position, and jumps to step s607; Step s605, PLC issues a command and the belt conveyor stops; Step s606, waiting for a manual signal, determining whether the belt conveyor is restarted, if restarted, jump to step s607, otherwise jump to step s609; Step s607, sending a power-on signal; Step s608, the belt conveyor starts running, and jumps to step s611; Step s609, no power-on signal is issued; Step s610, the belt conveyor stops and waits for manual maintenance, and jumps to step s611; Step s611, exit of the deviation control system sub-process.
2. The unattended conveyor system according to claim 1, characterized in that: The 1,000 collected image samples were manually labeled to mark the characteristics of tearing and deviation, and compared with the normal belt. The collected images were divided into training set: validation set: test set in the ratio of 7:2:1 using the YOLOv5 convolutional neural network. The model with weight YOLOv5s was trained for 500 rounds and a batch-size of 4. The training results were obtained and the model with the best training level was selected, which is the belt detection model.
3. The unattended conveyor system according to claim 1 is characterized in that: Step s404, determining whether the belt is deviating is as follows: The industrial camera obtains the normal working state picture of the belt, and converts the normal working state picture grayscale picture into a binary picture according to the set threshold value; sets the normal working area of the belt; Compare the real-time captured image binary image with the normal working state image binary image to determine whether the belt is deviating.
4. The unattended conveyor system according to claim 1, characterized in that: High-speed camera takes 30 photos at high speed.
5. The unattended conveyor system according to claim 1 is characterized in that: The SIFT corner point detection method is used to determine whether the belt is torn.
6. The unattended conveyor system according to claim 1, characterized in that: include: Central processing unit and inspection robot; The inspection robot includes: an intelligent detection platform, on which an industrial camera is set; The inspection robot is a hanging rail walking robot; The hanging rail of the hanging rail type walking robot is arranged above the belt conveyor.
Citation Information
Patent Citations
Machine vision belt tearing detection and protection device
CN116216230A
Centralized control system of adhesive tape machines
CN110077803A
Application of image recognition and AI algorithm in unattended operation of belt
CN114671215A
