Automatic snap ring intelligent identification system

Abstract

An automatic snap ring intelligent identification system comprises a data acquisition module, an automatic snap ring button detection model training module, an automatic snap ring button detection module, a state determination module and an alarm module, and is characterized in that the data acquisition module is provided with a camera; the automatic snap ring button detection model training module uses snap ring image data to label and train a YOLOV5-SEG model, the automatic snap ring button detection module triggers and collects images, the state determination module performs automatic snap ring button position determination, and the alarm module can perform sound-light alarm. The automatic snap ring intelligent identification system has the advantages of automatic monitoring, real-time monitoring, state prediction and diagnosis.

Description

Technical Field

[0001] The invention relates to an automatic snap ring of an automatic disassembly and assembly machine in a grinding roller room, in particular to an automatic snap ring intelligent identification system. Background Art

[0002] The TCM working roll and the middle roll can realize the automatic disassembly and assembly of the four-roll bearing seat at the same time. However, after the bearing seat is automatically installed, if the automatic retaining ring is not installed in place, it will affect the use of the roll in the rolling mill. In order to check and confirm whether the automatic retaining ring is installed in place, usually after the bearing seat is automatically installed in place, the automatic retaining ring is manually checked by camera to check whether it is installed in place. This inspection and confirmation method has high requirements for operators.

[0003] Therefore, in the prior art, how to provide a system for determining whether the automatic retaining ring of the automatic disassembly and assembly machine in the grinding roller room is installed in place has become a technical problem that needs to be solved urgently in this field. Summary of the invention

[0004] The purpose of the present invention is to provide a safe and reliable automatic clasp intelligent identification system.

[0005] To achieve the above purpose, the technical solution of the present invention is: An automatic clasp intelligent identification system includes a data acquisition module, an automatic clasp button detection model training module, an automatic clasp button detection module, a state determination module, and an alarm module, and is characterized in that: The data acquisition module is provided with a camera, and the camera selects a suitable camera type and the pixel requirements that meet the standards according to the operating environment and lighting conditions of the factory where the silicon steel roller shop disassembly and assembly machine is located; the number and location of the cameras are determined according to the monitoring range of the automatic inspection station of the roller shop disassembly and assembly machine; the data acquisition module selects a suitable network protocol and data transmission method, and adopts TCP / IP or Ethernet to ensure the stability and real-time performance of data transmission; The automatic snap ring button detection model training module uses snap ring image data for annotation, trains the YOLOV5-SEG model, and improves the accuracy and robustness of the model; it iterates and optimizes the model, optimizes the parameters of the YOLOV5-SEG model, and improves the convergence speed and performance of the model; uses perspective transformation, opening and closing operation OPENCV traditional algorithm to filter the mask image, and through Hough change post-processing operation, determines the accuracy and stability of whether the automatic snap ring is stuck; The automatic snap ring button detection module triggers image acquisition, collects pictures through the industrial ball camera, performs model reasoning, and transmits the collected real-time pictures to the segmentation algorithm model for reasoning, so that the deep learning algorithm model recognizes the mask area and two-dimensional coordinate information of the automatic snap ring button component, and obtains the world coordinate information of the automatic snap ring button component through the two-dimensional image calibration method; The state determination module determines the position of the automatic snap ring button. The system determines the position of the automatic snap ring button by the position of the automatic snap ring button fitting circle and the distance reference. If the automatic snap ring button fitting circle, the world coordinate system position and the distance between the two buttons are greater than the set threshold, it is determined that the automatic snap ring button is not fully opened. The alarm module can give an audible and visual alarm. When the automatic snap ring button is not fastened, the system can give an audible and visual alarm, send a text message or email to notify the operator that the equipment is in a stopped or abnormal state.

[0006] The automatic clasp intelligent identification system of the present invention can also be further implemented by adopting the following technical measures.

[0007] In the aforementioned automatic clip intelligent identification system, the camera is an industrial dome camera.

[0008] In the aforementioned automatic clasp intelligent identification system, the pixel of the industrial camera is more than 4 million.

[0009] In the aforementioned automatic clasp intelligent recognition system, the parameters of the YOLOV5-SEG model include learning rate and number of iterations.

[0010] In the aforementioned automatic clamp intelligent identification system, the threshold is a dynamic threshold, which can be dynamically adjusted according to different working conditions. The threshold can be appropriately adjusted for different types of round sticks.

[0011] After adopting the above technical solution, the automatic clasp intelligent identification system of the present invention has the following advantages: 1. Automated monitoring: Based on visual inspection technology, there is no need for manual inspection, which can realize automated monitoring, improve efficiency and reduce the impact of human factors; 2. Real-time monitoring: It can monitor the running status of the automatic retaining ring in real time, detect equipment abnormalities in time, and avoid equipment damage or product quality degradation; 3. Status prediction and diagnosis: Combine historical data and real-time monitoring data to predict possible problems of automatic clamping ring equipment in the future and achieve preventive maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 It is a schematic diagram of the principle diagram of the automatic clasp intelligent identification system according to an embodiment of the present invention.

[0013] In the figure: 1 automatic clip intelligent recognition system, 1.1 camera, 1.2 computing server, 1.3 HMI, 1.4 switch, 2 production PLC, 3 automatic clip. DETAILED DESCRIPTION

[0014] The present invention is further described below in conjunction with embodiments and accompanying drawings.

[0015] Example 1 Please refer to Figure 1 , Figure 1 The schematic diagram of the principle diagram of the automatic clasp intelligent identification system of the embodiment of the present invention is shown in FIG. The system framework of the automatic clasp intelligent identification system 1 of the present invention is composed of the following data acquisition module, automatic clasp button detection module, automatic clasp button detection module, state determination module, and alarm module: 1. Data acquisition module: ① Camera type: According to the operating environment and lighting conditions of the factory where the silicon steel roller shop disassembly and assembly machine is located, select the appropriate camera type and the pixel requirements that meet the standards, such as an industrial ball camera with more than 4 million pixels.

[0016] ② Number of cameras: Determine the number and position of cameras 1.1 according to the monitoring range of the automatic inspection station of the roller shop disassembly and assembly machine.

[0017] ③Data transmission method: Select appropriate network protocol and data transmission method, such as TCP / IP, Ethernet, etc., to ensure the stability and real-time performance of data transmission. Generally, independent TCP / IP protocol Ethernet is used.

[0018] 2. Automatic snap ring button detection model training module: ① Model training: Use the clasp image data with strong generalization for annotation and train the YOLOV5-SEG model to improve the accuracy and robustness of the model.

[0019] ② Model iteration and optimization: Optimize the parameters of the YOLOV5-SEG model, such as learning rate, number of iterations, etc., to improve the convergence speed and performance of the model.

[0020] ③ Image segmentation: Use OPENCV traditional algorithms such as perspective transformation and opening and closing operations to filter the mask image, and use post-processing operations such as Hough transformation to determine the accuracy and stability of the automatic clamping ring 3.

[0021] 3. Automatic snap ring button detection module: ①Trigger image acquisition: collect pictures through industrial ball camera.

[0022] ② Model reasoning: The collected real-time images are passed into the segmentation algorithm model for reasoning. The deep learning algorithm model identifies the mask area and two-dimensional coordinate information of the automatic snap ring button component, and obtains the world coordinate information of the automatic snap ring button component through the two-dimensional image calibration method.

[0023] 4. Status determination module: ① Automatic snap button position judgment: The system judges based on the position and distance of the automatic snap button fitting circle. If the automatic snap button fitting circle, the world coordinate system position, and the distance between the two buttons are greater than the set threshold, the automatic snap button is judged to be not fully opened.

[0024] ② Dynamic threshold: According to different working conditions, the threshold is adjusted dynamically. For different types of round sticks, the threshold can be adjusted appropriately.

[0025] 5. Alarm module: ① Sound and light alarm: When the automatic snap ring button is not fastened (when the position change of the "button" component is greater than the threshold), the system can issue an sound and light alarm, send a text message or email to notify the operator that the equipment is stopped or in an abnormal state.

[0026] The working process of the automatic clasp intelligent identification system of the present invention is as follows: This system uses a 4-megapixel 20x zoom dome camera with a focal length of 4mm. The dome camera can be rotated 360 degrees up and down, left and right, and the field of view of the dome camera can be adjusted within the range of 0 to 10m. The installation position of a dome camera can capture 4 rollers. The dome camera rotates at 4 angles, and captures a roller at each rotation angle. The installation range of the dome camera from the roller is estimated to be between 3 and 5m, depending on the specific situation on site. Each dome camera is equipped with an LED light source, and the light source switch is on the console in the central control room. The staff determines whether to turn on the light according to the on-site environment.

[0027] The automatic snap ring image captured by the ball camera is transmitted to the visual computing server 1.2, the image is analyzed by the visual detection algorithm, and combined with the image recognition algorithm, it is determined whether the automatic snap ring is in an open state or a closed state.

[0028] The targets are calibrated on different inner ring areas of the roller, and through the visual algorithm processing part, four targets of different diameters are calibrated on the side of the roller.

[0029] The final result is displayed through the HMI1.3 terminal and voice prompts are given.

[0030] The present invention has substantial characteristics and significant technical progress. The automatic clasp intelligent identification system of the present invention is aimed at the following traditional technologies: relying on manual inspection and naked eye viewing, which is inefficient and has the risk of personal injury and is easily affected by human factors; lacking real-time monitoring, and failing to detect equipment abnormalities in time, which may cause damage to the automatic clasp equipment or deterioration of product quality; failing to predict and diagnose the equipment operation status, and failing to achieve preventive maintenance. It realizes automatic monitoring: based on visual detection technology, no manual inspection is required, automatic monitoring is realized, efficiency is improved, and the influence of human factors is reduced. Real-time monitoring is realized, and the operation status of the automatic clasp can be monitored in real time, and equipment abnormalities can be discovered in time to avoid equipment damage or product quality deterioration. State prediction and diagnosis: combining historical data and real-time monitoring data, predicting possible problems of the automatic clasp equipment in the future, and achieving preventive maintenance; analyzing the cause of equipment failure, and providing maintenance suggestions. Improving the level of production automation: the automatic clasp intelligent identification system of the present invention combines visual detection technology with deep learning algorithms, realizes real-time monitoring and intelligent analysis of the buckle status of the automatic clasp, and greatly improves the level of production automation.

[0031] The automatic clasp intelligent identification system of the present invention is an automatic clasp buckle detection solution based on visual detection, which can accurately judge the operating status of the equipment, avoid equipment damage and reduce production efficiency. The system has the characteristics of high efficiency, intelligence and scalability through real-time data collection and analysis, greatly improving the level of production automation.

[0032] The scalability and optimization of the automatic clasp intelligent identification system of the present invention are shown in the following aspects: 1. Equipment upgrade and multi-sensor fusion: 1) 3D laser line scanning: Using the visual inspection results after 3D laser line scanning to understand the depth information of the automatic buckle, you can more comprehensively evaluate the operating status of the equipment.

[0033] 2) Vibration sensor: By monitoring the automatic retaining ring button, it can determine whether there is abnormal vibration, thereby detecting potential faults earlier.

[0034] 3) Pressure sensor: detects the pressure change of the automatic clamping ring button and finds problems such as failure to lock in time.

[0035] 2. Deep learning model improvements: 1) Introducing the attention mechanism: focusing on key components such as the automatic snap ring button to improve the model’s robustness to occlusion, lighting changes, etc.

[0036] 2) Combined with transfer learning: Use the pre-trained model for fine-tuning to adapt to the detection needs of different models of automatic snap ring buttons.

[0037] 3) Try other models: such as EfficientDet, SSD, etc., to find a model that is more suitable for this detection task.

[0038] 3. System function expansion: 1) Equipment status prediction: Based on historical data and real-time monitoring data, predict possible problems of automatic snap ring buttons in the future to achieve preventive maintenance.

[0039] 2) Equipment fault diagnosis: Combine sensor data and visual inspection results to analyze the cause of equipment failure and provide maintenance suggestions.

[0040] 3) Production process optimization: According to the operating status of the automatic retaining ring, adjust the production parameters to improve production efficiency and product quality.

[0041] 4. Improved system security: 1) Data encryption: Encrypt the collected data for storage and transmission to prevent data leakage.

[0042] 2) Permission management: Set different access rights for different users to ensure system security.

[0043] 3) System backup: Back up system data regularly to prevent data loss.

[0044] Through the above expansion and optimization, the automatic snap ring button recognition system will be more intelligent, reliable and easy to use, providing more efficient and safer production processes in the steel industry.

[0045] The above embodiments are only for the purpose of illustrating the present invention, but not for limiting the present invention. A person skilled in the art may make various changes or modifications without departing from the spirit and scope of the present invention. Therefore, all equivalent technical solutions should also belong to the scope of the present invention and should be defined by the claims.

Claims

1. An automatic clasp intelligent identification system, comprising a data acquisition module, an automatic clasp button detection model training module, an automatic clasp button detection module, a state determination module, and an alarm module, characterized in that: The data acquisition module is provided with a camera, which selects a suitable camera type and pixel requirements according to the operating environment and lighting conditions of the factory where the silicon steel roller shop disassembly and assembly machine is located; and determines the number and position of the camera according to the monitoring range of the automatic inspection station of the roller shop disassembly and assembly machine; The data acquisition module selects appropriate network protocols and data transmission methods, using TCP / IP or Ethernet to ensure the stability and real-time performance of data transmission; The automatic snap ring button detection model training module uses snap ring image data for annotation, trains the YOLOV5-SEG model, and improves the accuracy and robustness of the model; it iterates and optimizes the model, optimizes the parameters of the YOLOV5-SEG model, and improves the convergence speed and performance of the model; uses perspective transformation and opening and closing operation OPENCV traditional algorithms to filter the mask image, and performs Hough change post-processing operations to determine the accuracy and stability of whether the automatic snap ring is stuck; The automatic snap ring button detection module triggers image acquisition, collects pictures through the industrial ball camera, performs model reasoning, and transmits the collected real-time pictures to the segmentation algorithm model for reasoning, so that the deep learning algorithm model recognizes the mask area and two-dimensional coordinate information of the automatic snap ring button component, and obtains the world coordinate information of the automatic snap ring button component through the two-dimensional image calibration method; The state determination module determines the position of the automatic snap ring button. The system determines the position of the automatic snap ring button by the position of the automatic snap ring button fitting circle and the distance reference. If the automatic snap ring button fitting circle, the world coordinate system position and the distance between the two buttons are greater than the set threshold, it is determined that the automatic snap ring button is not fully opened. The alarm module can give an audible and visual alarm. When the automatic snap ring button is not fastened, the system can give an audible and visual alarm, send a text message or email to notify the operator that the equipment is in a stopped or abnormal state.

2. The automatic clasp intelligent identification system according to claim 1, characterized in that: The camera is an industrial dome camera.

3. The automatic clasp intelligent identification system according to claim 1, characterized in that: The industrial camera has a pixel of more than 4 million.

4. The automatic clasp intelligent identification system according to claim 1, characterized in that: The parameters of the YOLOV5-SEG model include learning rate and number of iterations.

5. The automatic clasp intelligent identification system according to claim 1, characterized in that: The threshold is a dynamic threshold, which can be dynamically adjusted according to different working conditions. The threshold can be appropriately adjusted for different types of round sticks.

Images