A steel mill plug detection system

Abstract

The present invention discloses a steel mill plug detection system, comprising a position detection trigger device, a high-speed camera device, an industrial control host, and an alarm device; the position detection trigger device, the high-speed camera device, and the alarm device are all communicatively connected to the industrial control host; the position detection trigger device is arranged at the front end of the high-speed camera device; the high-speed camera device, in response to a start signal sent by the industrial control host, continuously shoots a preset position at high speed with preset shooting parameters, and transmits the obtained n pictures to the industrial control host; a plug feature recognition model trained in the industrial control host recognizes the received pictures and gives a conclusion on whether the plug feature exists in the pictures; if the plug feature does not exist in the pictures, the industrial control host sends a trigger signal to the alarm device; and the alarm device issues an alarm message to the outside according to the trigger signal from the industrial control host. Thus, the steel mill plug detection system of the present invention can completely replace manual inspection of the presence of plugs.

Description

Technical Field

[0001] The present invention relates to the field of steel pipe production and processing, and in particular to a steel plant plug detection system. Background Art

[0002] PQF limited mandrel continuous pipe rolling technology is a pipe rolling process technology developed in the mid-1990s. It has made great progress in the 21st century and represents the latest level of process design and manufacturing of seamless steel pipe rolling mills today. The piercing machine is the key equipment of the PQF continuous rolling mill. It is a spiral piercing structure, mainly composed of a piercing body, pre-rotation of the front-end tube billet, three-roller centering in the back-end, and a push rod trolley. The push rod adopts a replaceable plug structure, and the plug is removed by the plug replacement box to achieve plug cooling and recycling. In daily production, due to wear on the joint between the plug and the push rod, failure of the plug device, failure of the plug locking device, etc., the plug may be detached from the correct position, and the plug may fall off when the jacking pipe carries the plug or during the rolling process. If the plug falls into the shell tube cavity and the operator fails to detect it in time, it will enter the PQF continuous rolling mill with the shell tube, causing the continuous rolling mill to jam or pile up steel, and even damage the rolls, couplings, and shafts. If the plug falls during the forward movement and is not discovered in time, the piercer will roll without the plug, causing damage to the ejector rod and the three-roll centering device. This can lead to serious consequences such as long production line downtime. The shutdown of a production line can force the suspension of previous and subsequent processes, resulting in significant economic losses. Currently, the only way to ensure the correct positioning of the plug is to manually observe the front end of the piercer. However, the working environment at the front end of the piercer is harsh, with high temperature, high humidity, and high noise. It is also very dangerous. Moreover, it relies on workers' visual inspection, and the results are unstable. Currently, similar PQF units in China do not have a good solution for detecting the correct positioning of the plug. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a steel plant plug detection system.

[0004] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0005] A steel mill plug detection system includes a position detection trigger device, a high-speed camera device, an industrial control host, and an alarm device; the position detection trigger device, the high-speed camera device, and the alarm device are all communicatively connected to the industrial control host;

[0006] The position detection trigger device is arranged at the front end of the high-speed camera device, and is used to detect whether there is an object passing along the expected movement trajectory of the ram. When an object is detected passing, a position trigger signal is transmitted to the industrial control host.

[0007] After receiving a start signal from the industrial control host, the high-speed camera device continuously shoots at a preset position with preset shooting parameters at high speed, and transmits the obtained n pictures to the industrial control host;

[0008] The industrial control host is equipped with a trained head feature recognition model, which recognizes the received image and determines whether the head feature exists in the image; if the head feature does not exist in the image, the industrial control host sends a trigger signal to the alarm device;

[0009] The alarm device sends out an alarm message according to the trigger signal of the industrial control host.

[0010] Compared with the prior art, the present invention has the following technical effects:

[0011] After the position detection trigger device detects that a moving object has entered the test area, it sends a position trigger signal to the industrial control host. After receiving the position trigger signal, the industrial control host sends a start signal to the high-speed camera device. After receiving the start signal, the high-speed camera device immediately performs high-speed continuous shooting according to the preset shooting parameters to obtain multiple pictures. By reasonably arranging the position of the position detection trigger device and telling the camera device, and setting the shooting parameters, theoretically, all the photos taken should contain the head image. These photos are sent to the industrial control host. The built-in head feature recognition module of the industrial control host searches for the head feature in each photo. If the preset head feature is contained in the photos of the set proportion, the head is considered normal, otherwise the head is considered abnormal. The head abnormality includes the situation of complete detachment and the situation of about to fall. Which situation is normal and which situation is abnormal can be formed when the head feature recognition model is trained. As long as enough photos of normal conditions and abnormal conditions are provided, the head feature recognition model can learn the recognition function.

[0012] On the basis of the above technical solution, the present invention can also be improved as follows.

[0013] Preferably, the position detection trigger device is communicatively connected to a control device that controls the movement of the ram, and the control device sends a ram-about-arrival signal to the trigger device. After receiving the ram-about-arrival signal, the position detection trigger device starts the detection work and stops the detection after detecting a moving object. Thus, the position detection trigger device only starts the work when the ram is about to arrive and stops the work after detecting a moving object, thereby reducing the probability of false triggering.

[0014] Preferably, it also includes a speed measuring device, which is arranged in front of the position detection trigger device, and is used to detect the movement speed v of the object on the expected movement trajectory of the ram, and transmit the detection result to the industrial control host. The industrial control host adjusts the time difference between the position trigger signal and the high-speed camera device start signal based on the speed v and the distance between the position detection trigger device and the high-speed camera device. Because there is a certain distance between the position detection trigger device and the high-speed camera device, there must be a certain time difference from the moment the position detection trigger device detects a moving object to the moment the moving object enters the camera range of the high-speed camera device, and this time difference will change with the movement speed of the moving object. After adding the speed measuring device, this speed value can be used to adjust the time difference between the position trigger signal and the high-speed camera device start signal to ensure that after the high-speed camera device is started, the moving object just passes through the shooting range of the high-speed camera device, thereby ensuring that the photos taken all contain the expected position of the ram. Therefore, this device can be applied to production lines with different movement speeds.

[0015] Preferably, if the plug feature is absent from the image, the industrial control host computer sends a linkage signal to the control device controlling the plug's movement, causing it to stop immediately. When multiple photos are available, a plug abnormality is determined only when the plug feature is absent from a set ratio of photos. Immediately stopping the plug's movement can prevent the accident from escalating and minimize losses.

[0016] Preferably, the shooting parameters include exposure mode, continuous shooting interval, etc., which are adjusted according to the actual situation on site, the speed of the ejector, the lighting conditions, etc., to ensure that photos that meet the requirements are obtained.

[0017] Preferably, an image preprocessing module is provided in the industrial control host, which converts the image into a black and white image, removes small spots in the photo, because these small spots are generally interference from water droplets and dust in the environment, and performs appropriate morphological processing, such as strengthening the boundary line, etc., to highlight the head features.

[0018] Preferably, the plug feature recognition model is based on deep learning of convolutional neural networks. As production inspection proceeds, pictures of plugs and plug drops under different working conditions are continuously collected, the training library of deep learning is gradually increased, and iterative training is performed to steadily improve the accuracy of the optimized detection system.

[0019] Preferably, the high-speed camera device is arranged in a protective shell, and a vortex cold air duct is provided in the protective shell to continuously cool the high-speed camera device, and a compressed air nozzle is provided outside the protective shell. The compressed air nozzle forms a vortex air curtain outside the protective shell to prevent dust and water vapor from contaminating the lens and camera.

[0020] Preferably, the speed measuring device is a Doppler speed measuring radar, which can accurately measure the speed of an object running at a speed of 4-5 meters per second.

[0021] Preferably, the position detection trigger device is a laser sensor with a measurement accuracy of up to 0.5 mm, a measurement frequency of 400 Hz, and an output frequency of 1200 Hz, which can accurately sense a moving centerline speed of 4-5 m / s. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 This is a schematic diagram of the electrical structure of the steel mill plug detection system of the present invention;

[0023] Figure 2 This is a schematic diagram of the installation of some components of the steel mill plug detection system of the present invention;

[0024] Figure 3 、 Figure 4 、 Figure 5 Schematic diagram of three different states of the detection process of the steel mill plug detection system of the present invention;

[0025] Figure 6 Shown is a photo of the detected head and the passing result displayed on the display screen;

[0026] Figure 7 Shown are photos of possible missing pins and failed results displayed on the display screen;

[0027] Figure 8 Shown is a photo of a situation where the top head has completely fallen off;

[0028] Figure 9 Shown is a photo of a plug about to fall off.

[0029] In the accompanying drawings, the names of the components represented by the reference numerals are listed as follows:

[0030] 1. Heavy-duty support frame; 2. High-speed camera; 3. Laser sensor; 4. Doppler speed radar; 5. Protective housing; 6. Industrial control host; 7. Display; 8. Alarm speaker; 9. PLC; 10. Header. DETAILED DESCRIPTION

[0031] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only used to explain the present invention and are not used to limit the scope of the present invention.

[0032] Please refer to Figure 1As shown, it is a structural schematic diagram of the steel mill head 10 detection system of the present invention. The steel mill head 10 detection system includes: a heavy-duty support frame 1, a high-speed camera 2, a laser sensor 3, a Doppler speed radar 4, a protective housing 5, an eddy current cooling air duct, an industrial control host 6 embedded with deep learning and machine vision algorithms, a display 7, and an alarm speaker 8. The laser sensor 3 is a position detection trigger device, and the high-speed camera 2 is a high-speed camera device. The position detection trigger device, high-speed camera device, and alarm device are all connected to the industrial control host 6 via fiber optic communication. The high-speed camera used in this detection system uses Gigabit Ethernet to convert Gigabit fiber optic transmission images. The Ethernet cable and optical fiber are equipped with a shielding layer, which has the characteristics of fast transmission speed, long transmission distance, and strong anti-interference ability. The display screen displays the images captured by the high-speed camera 2 in real time. The high-speed camera 2 and laser sensor 3 are installed on the heavy-duty support frame 1.

[0033] The laser sensor 3 is provided at the front end of the high-speed camera 2 and is used to detect whether there is an object (the mandrel 10 or the mandrel) passing along the expected motion trajectory of the mandrel 10. When an object is detected passing, a position trigger signal is transmitted to the industrial control host 6. In order to avoid unnecessary false triggering, the laser sensor 3 is connected to the PLC 9 that controls the movement of the mandrel 10. Figure 3 As shown, when the mandrel 10 is far away from the detection position, the laser sensor 3 and the high-speed camera 2 are both in the off state, as shown in FIG. Figure 4 As shown, when the plug 10 is about to reach the detection position, the PLC 9 will send a signal that the plug 10 is about to arrive to the trigger device. After the laser sensor 3 receives the signal that the plug 10 is about to arrive, the laser sensor 3 starts the detection work, and when a moving object is detected, it sends a signal and stops the detection. After the industrial control host 6 receives the signal that the moving object is detected from the laser sensor 3, it sends a start signal to the high-speed camera 2. Figure 5 As shown, the high-speed camera 2 starts filming the expected plug 10 passing by. (If the plug 10 falls, the ejector rod will pass in front of the high-speed camera 2.) The laser sensor 3 used in this detection system has a measurement accuracy of up to 0.5mm, a measurement frequency of 400Hz, and an output frequency of 1200Hz. It can accurately sense the moving plug 10 with a centerline velocity of 4-5 m / s.

[0034] The Doppler velocity radar 4, located in front of the laser sensor 3, is used to detect the velocity v of objects along the expected trajectory of the plug 10 and transmit the detection results to the industrial control host 6. The industrial control host 6 adjusts the time difference between the position trigger signal and the high-speed camera start signal based on the velocity v and the distance between the laser sensor 3 and the high-speed camera 2. Thus, the Doppler velocity radar 4 measures the velocity of the plug 10 in real time, and calculates the appropriate imaging interval based on the velocity. This system is compatible with plugs 10 of varying linear speeds and lengths and diameters, ensuring accurate capture of the plug 10's features. This system can detect plugs 10 with linear speeds of up to 4-5 m / s and is compatible with plugs 10 of varying lengths and diameters.

[0035] This inspection system uses an industrial camera with a frame rate of up to 600fps and global shutter exposure, enabling clear capture of the moving centerline speed of 4-5 m / s. Following a start signal from the industrial control host 6, the high-speed camera 2 continuously captures images at a preset location using preset shooting parameters at high speed, and transmits the six images obtained to the industrial control host 6. The high-speed camera 2 is housed within a protective housing 5, which is equipped with a vortex cooling air duct to continuously cool the high-speed camera 2. Furthermore, a compressed air nozzle is provided outside the protective housing 5, forming a vortex-shaped air curtain outside the protective housing 5. With this protection, the system can adapt to harsh, complex, and changing production environments, and the hardware used can withstand high temperature, high humidity, high dust content, and easily contaminated production environments.

[0036] The industrial control host 6 is provided with an image preprocessing module, which converts the image into a black and white image and removes small spots in the photo. The small spots in the image are generally interference caused by on-site dust, water droplets, etc., which need to be optimized and removed. The image can also be subjected to enhanced boundary line processing to highlight the features of the plug 10; the industrial control host 6 is provided with a trained plug 10 feature recognition model, which recognizes the received image and gives a conclusion on whether the plug 10 feature exists in the image; if the plug 10 feature does not exist in the image, the industrial control host 6 sends a trigger signal to the alarm device; at the same time, it sends a linkage signal to the PLC 9 that controls the movement of the plug 10 to stop the movement of the plug 10 urgently.

[0037] The plug 10 feature recognition model is based on deep learning using a convolutional neural network. As production inspections progress, images of the plug 10 under various operating conditions and images of dropped plugs 10 are continuously collected, gradually expanding the deep learning training library and steadily improving the accuracy of the optimized inspection system. This online inspection system, based on deep learning and machine vision, utilizes the CPU of the industrial control host 6 for machine vision and the GPU of the industrial control host 6 for deep learning. This dual-channel processing delivers high computational speed, enabling the system to complete image capture, inspection, and result output within one second. The system's display screen displays an operational interface that displays images of detected plugs and inspection results.

[0038] Figure 6 Shown is a photo of the detected head and the passing result displayed on the display.

[0039] Figure 7 Shown are photos of possible missing pins and the failed result displayed on the display.

[0040] Figure 8 Shown is a photo of a situation where the top head has completely fallen off;

[0041] Figure 9 Shown is a photo of the situation where the top head is about to fall off;

[0042] The alarm device, i.e., the alarm speaker, sends out an alarm message according to the trigger signal of the industrial control host 6.

[0043] The pictures taken by the high-speed camera 2 can be stored in a set location to facilitate the subsequent tracing of previous inspection results and to provide training materials for the feature recognition model of the plug 10.

[0044] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A steel mill plug detection system, characterized in that: It includes a position detection trigger device, a high-speed camera device, an industrial control host and an alarm device; the position detection trigger device, the high-speed camera device and the alarm device are all connected to the industrial control host in communication; The position detection trigger device is arranged at the front end of the high-speed camera device, and is used to detect whether there is an object passing along the expected movement trajectory of the ram. When an object is detected passing, a position trigger signal is transmitted to the industrial control host. After receiving a start signal from the industrial control host, the high-speed camera device continuously shoots at a preset position with preset shooting parameters at high speed, and transmits the obtained n pictures to the industrial control host; The industrial control host is equipped with a trained head feature recognition model, which recognizes the received image and determines whether the head feature exists in the image; if the head feature does not exist in the image, the industrial control host sends a trigger signal to the alarm device; The alarm device sends out an alarm message according to the trigger signal of the industrial control host; The position detection trigger device is a laser sensor; The laser sensor is connected to the PLC that controls the movement of the ram. When the ram is far away from the detection position, the laser sensor and the high-speed camera are both in the off state. When the ram is about to reach the detection position, the industrial control host sends a ram arrival signal to the trigger device. After receiving the ram arrival signal, the laser sensor starts the detection work, and when a moving object is detected, it sends a signal and stops the detection.

2. The steel mill plug detection system according to claim 1, characterized in that: The position detection trigger device is in communication with the control device for controlling the movement of the plug. The control device sends a plug arrival signal to the trigger device. After receiving the plug arrival signal, the position detection trigger device starts the detection work and stops the detection after detecting a moving object.

3. The steel mill plug detection system according to claim 1, characterized in that: It also includes a speed measuring device, which is located in front of the position detection trigger device and is used to detect the movement speed of the object on the expected movement trajectory of the head. v and transmit the detection results to the industrial control host, which v , combined with the distance between the position detection trigger device and the high-speed camera device, adjust the time difference between the position trigger signal and the high-speed camera device start signal.

4. The steel mill plug detection system according to claim 1, characterized in that: If there is no ram feature in the image, the industrial control host sends a linkage signal to the control device that controls the movement of the ram to urgently stop the movement of the ram.

5. The steel mill plug detection system according to claim 1, characterized in that: The shooting parameters include exposure mode and continuous shooting interval.

6. The steel mill plug detection system according to claim 1, characterized in that: The industrial control host is provided with an image pre-processing module, which converts the image into a black and white image, removes small spots in the photo, and strengthens the boundary line to highlight the top head feature.

7. The steel mill plug detection system according to claim 1, characterized in that: The plug feature recognition model is based on deep learning of convolutional neural networks. As production inspection progresses, pictures of plugs and plug drops under different working conditions are continuously collected, the deep learning training library is gradually increased, and the accuracy of the optimized inspection system is steadily improved.

8. The steel mill plug detection system according to claim 1, characterized in that: The high-speed camera device is arranged in a protective shell, and a vortex cold air duct is provided in the protective shell to continuously cool the high-speed camera device. A compressed air nozzle is provided outside the protective shell to form a vortex-shaped air curtain outside the protective shell.

9. The steel mill plug detection system according to claim 3, characterized in that: The speed measuring device is a Doppler speed measuring radar.

Images