A safety production supervision early warning method and system based on AI recognition
Through AI recognition technology, real-time monitoring of production workshop videos and identifying the center of mass trajectory of moving changes areas, solving the problem of lack of real-time safety monitoring in the existing technology, improving the accuracy and reliability of production safety, reducing manual intervention, and achieving the goal of Comprehensive monitoring and early warning of the production process.
Patent Information
- Application Number
- CN202410569759.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-09
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2044-05-09
AI Technical Summary
The existing production safety monitoring methods mainly rely on manual alarms and lack real-time safety monitoring systems, which can affect product production quality when equipment failure or personnel operation is not standardized and may cause personal safety accidents.
Using AI-based safety supervision and early warning method, by obtaining monitoring instructions, obtaining production workshop videos, performing grayscale processing and preset dividing lines, dividing work areas, calculating the center of mass trajectory of the moving change area, and determining whether it is periodic motion , If otherwise it is determined to be an abnormal situation, prompt management and security personnel.
Real-time monitoring of the production process is achieved, the efficiency and accuracy of safe production is improved, manual intervention is reduced, the misjudgment rate is reduced, abnormal situations can be monitored more comprehensively, and the reliability of early warning is improved.
Smart Images

Figure CN118505624B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of safe production technology, and in particular to a safe production supervision and early warning method and system based on AI recognition. Background Art
[0002] In today's highly industrialized society, assembly line production has become the most common and widely used production method in modern manufacturing. In the assembly line production process, high-intensity operation of machinery and equipment is inevitable. Long-term continuous work will cause equipment wear, aging, and even failure. Once a machine fails, it will not only affect the production progress, but may also cause accidents and cause personal injury. In addition, equipment failure may also lead to problems such as material leakage and stacking during the production process, increasing the risk of accidents such as fire and explosion.
[0003] Secondly, during the production process of the assembly line, human operation errors are also a major safety hazard, such as failure to wear protective equipment as required, accidentally touching the switch, etc. These errors may lead to equipment loss of control, material blockage, etc., which in turn may cause safety accidents.
[0004] Therefore, when production equipment fails or personnel operate improperly, it will affect product quality and even cause personal safety accidents. Existing production safety monitoring methods mainly rely on manual alarms and lack a real-time safety monitoring system. Summary of the invention
[0005] The purpose of the present invention is to provide a safety production supervision early warning method and system based on AI recognition to solve the following technical problems:
[0006] When production equipment fails or personnel operate improperly, it will affect product quality and even cause personal safety accidents. Existing production safety monitoring methods mainly rely on manual alarms and lack a real-time safety monitoring system.
[0007] The purpose of the present invention can be achieved by the following technical solutions:
[0008] A safety production supervision and early warning method based on AI recognition includes the following steps:
[0009] Obtain monitoring instructions, which include but are not limited to electrical signals, motion signals and voice signals, and which include machine start instructions and machine stop instructions;
[0010] When a machine start instruction is collected, a monitoring video in the production workshop is obtained, split to obtain a monitoring image sequence, and grayscale processing is performed on the monitoring image sequence to generate a grayscale image sequence;
[0011] Identify a preset dividing line in a grayscale image, the preset dividing line is actually set in the production workshop, segment a plurality of working area images from any grayscale image according to the preset dividing line, establish a rectangular coordinate system with any fixed point of the working area image as the origin, identify the working area image, and divide the working area image into a background area and a change area according to the change of the grayscale value of the pixel point;
[0012] According to the coordinates of each pixel point in the change area in the grayscale image sequence, the pixel points that are always in the change area are divided into the static change area, and the pixel points that are intermittently in the change area are divided into the motion change area;
[0013] For any motion change area in the image of the working area, the coordinates of all its pixel points are obtained, and the motion change area is equivalent to a planar entity with uniform density. The center of mass coordinates of the motion change area are calculated according to the coordinates of all its pixel points, and a motion trajectory of the center of mass coordinates changing with time is generated. It is determined whether the motion trajectory is a periodic identical motion. If so, monitoring continues; if not, and no machine stop command is collected at this time, it is determined to be an abnormal situation, and the management personnel and safety personnel in the production workshop are prompted.
[0014] As a further solution of the present invention: when the monitoring instruction is an electrical signal or a voice signal, it is collected by an electrical signal acquisition module or a voice recognition module arranged in the production equipment; when the monitoring instruction is an action signal, it is collected by an image recognition module in the workshop monitoring camera.
[0015] As a further solution of the present invention: the color or shape of the preset dividing line is different from the workshop background, and the color or shape of the preset dividing line is entered into the image recognition module.
[0016] As a further solution of the present invention: the process of dividing the background area and the change area is:
[0017] Obtain the grayscale values (xi, yi) of all pixels of all grayscale images in the grayscale image sequence, where x and y represent the coordinates of any pixel in the grayscale image, i represents the time sequence of the grayscale image where the pixel is located in the entire grayscale image sequence, and i is a positive integer. By calculating the grayscale difference d of the pixels with the same coordinates between the grayscale images at adjacent times, if the grayscale difference d is greater than a preset illumination error threshold, the pixels with the same coordinates are the change area; if the grayscale difference d is less than the preset illumination error threshold, the pixels with the same coordinates are the background area.
[0018] As a further solution of the present invention: the specific process of determining the static change area and the motion change area is:
[0019] Get the change area in the time series, collect the coordinates of the pixel points in the change area of each frame in turn, if there is a pixel point with any coordinates that is always in the change area, then the pixel point is determined to be a static change area; if there is a pixel point with any coordinates that belongs to both the change area and the background area in the time series, then the pixel point is a moving change area.
[0020] As a further solution of the present invention: during the monitoring process, if the shape or size of the motion trajectory changes and no machine stop instruction is collected, it is still determined to be an abnormal situation even if the motion trajectory after the change is still a periodic and identical motion.
[0021] As a further solution of the present invention: during the monitoring process, if no machine stop instruction is collected, if the pixel area of the motion change area decreases to less than a% of the original pixel area, a is a preset value, it is also determined to be an abnormal situation.
[0022] A safety production supervision and early warning system based on AI recognition, including:
[0023] A signal synchronization module, used to obtain monitoring instructions, including but not limited to electrical signals, action signals and voice signals, and including machine start instructions and machine stop instructions;
[0024] An image acquisition module is used to acquire the monitoring video in the production workshop when a machine start instruction is collected, split it to obtain a monitoring image sequence, and perform grayscale processing on the monitoring image sequence to generate a grayscale image sequence;
[0025] A change recognition module is used to recognize a preset dividing line in a grayscale image, wherein the preset dividing line is actually set in a production workshop, and a plurality of working area images are segmented from any grayscale image according to the preset dividing line, and a rectangular coordinate system is established with any fixed point of the working area image as the origin, and the working area image is recognized, and the working area image is divided into a background area and a change area according to the change of the grayscale value of the pixel point;
[0026] A motion recognition module is used to classify the pixels that are always in the changing area into the static changing area and the pixels that are intermittently in the changing area into the moving changing area according to the coordinates of each pixel in the changing area in the grayscale image sequence;
[0027] The trajectory determination module is used to obtain the coordinates of all pixel points in the motion change area in any working area image, regard the motion change area as equivalent to a plane entity with uniform density, calculate the center of mass coordinates of the motion change area according to the coordinates of all pixel points, generate a motion trajectory of the center of mass coordinates changing with time, and determine whether the motion trajectory is a periodic identical motion. If so, continue monitoring; if not, and no machine stop command is collected at this time, it is determined to be an abnormal situation, and the management personnel and safety personnel in the production workshop are prompted.
[0028] Beneficial effects of the present invention:
[0029] The present invention utilizes the characteristics of periodic mechanical motion of production equipment and operators during assembly line operation, monitors the working area where the production equipment is located through monitoring images, identifies the motion trajectory of the centroid of the motion change area through grayscale processing, pixel point coordinates and equivalent entities, and determines whether the motion trajectory is a periodic identical motion. It can further determine whether the production process is proceeding normally, thereby improving the accuracy of recognition. When the shape or size of the motion trajectory changes, it is also determined to be an abnormal situation. It can more comprehensively monitor abnormal situations in the production process and improve the reliability of recognition. BRIEF DESCRIPTION OF THE DRAWINGS
[0030] The present invention will be further described below in conjunction with the accompanying drawings.
[0031] Figure 1 It is a module schematic diagram of a production safety supervision and early warning system based on AI recognition in the present invention. DETAILED DESCRIPTION
[0032] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0033] See also Figure 1 As shown, the present invention is a safety production supervision and early warning method based on AI recognition, comprising the following steps:
[0034] Obtain monitoring instructions, which include but are not limited to electrical signals (buttons), action signals (body movements) and voice signals (commands), and the monitoring instructions include machine start instructions and machine stop instructions; when the monitoring instructions are electrical signals or voice signals, that is, when the operator controls the equipment through buttons or voice, the monitoring instructions are collected through an electrical signal acquisition module or a voice recognition module arranged in the production equipment; when the monitoring instructions are action signals, that is, the equipment requires large body movements by the operator to operate, the monitoring instructions can be collected through the image recognition module in the workshop monitoring camera.
[0035] When a monitoring instruction for starting the machine is collected, the monitoring video captured by the monitoring camera in the production workshop is retrieved in real time, the generated monitoring video is split into a monitoring image sequence, and the monitoring image sequence is gray-processed to generate a gray-scale image sequence;
[0036] Identify a preset dividing line in the grayscale image, which is a landmark line used to distinguish the equipment area and the aisle area in the factory workshop. The landmark line is actually set in the production workshop. The color or shape of the landmark line is different from the workshop background. The color or shape of the landmark line is entered into the image recognition module. According to the landmark line, a working area image corresponding to the equipment area is segmented from any grayscale image;
[0037] A rectangular coordinate system is established with any fixed point of the working area image as the origin, and the working area image is identified; the grayscale values (xi, yi) of all pixels of all grayscale images in the grayscale image sequence are obtained, where x and y represent the coordinates of any pixel in the grayscale image, i represents the time sequence of the grayscale image where the pixel is located in the entire grayscale image sequence, and i is a positive integer. The grayscale difference d of the same coordinate pixels between the grayscale images at adjacent times is calculated, d=|G(x i+1 ,y i+1 )-G(x i ,y i )|, if the grayscale difference d is greater than the preset illumination error threshold (the illumination error threshold is the illumination change caused by time change within the time difference corresponding to adjacent image frames, and is set according to the actual situation), then the pixel point with the same coordinates is the change area; if the grayscale difference is less than the preset illumination error threshold, then the pixel point with the same coordinates is the background area.
[0038] According to the coordinates of each pixel point in the change area in the grayscale image sequence, the pixel points that are always in the change area are divided into the static change area, and the pixel points that are intermittently in the change area are divided into the motion change area; the specific process of determining the static change area and the motion change area is:
[0039] Get the changing area in the time series, collect the coordinates of the pixel points in the changing area of each frame in turn, if there is a pixel point with any coordinates that is always in the changing area, then the pixel point is determined to be a static changing area; if there is a pixel point with any coordinates that belongs to both the changing area and the background area in the time series, then the pixel point is a moving changing area. For a person whose body is still and only his hands move, the static changing area is his body, and the moving changing area is his arm. For a machine, the static changing area is the conveyor belt and operating table, etc., and the moving changing area is the robotic arm and other moving parts.
[0040] For any motion change area in the image of the working area, the coordinates of all its pixel points are obtained, and the motion change area is equivalent to a plane entity with uniform density. According to the coordinates of all pixel points in the plane, the shape of the plane entity is generated, so as to directly calculate the center of mass of the plane entity (i.e., the plane sheet), which is the center of mass coordinates of the motion change area, and generate a motion trajectory of the center of mass coordinates changing with time. The motion trajectory is the line connecting the center of mass coordinates in continuous time, and it is judged whether the motion trajectory is a periodic identical motion. If so, it indicates the regularity of the motion trajectory, and the operator and the machine are repeating the motion, which means that the operator and the equipment are in normal production activities, and the monitoring continues; if not, and the machine stop command is not collected at this time, it means that the motion of the machine or the personnel has changed, which may be a machine failure, or the personnel may have physical abnormalities or leave the post without authorization, then it is determined to be an abnormal situation, and the management personnel and safety personnel in the production workshop are prompted to check the working area of the equipment.
[0041] The present invention can timely detect abnormal situations in the production process by real-time monitoring of machine start-up instructions and surveillance videos, thereby improving the efficiency and accuracy of production safety supervision. By using AI technology to perform grayscale processing on the monitoring image sequence, identify preset dividing lines, and divide the change area, the motion change area in the working area can be automatically identified, reducing manual intervention and lowering the misjudgment rate. The motion trajectory is generated based on the centroid coordinates of the motion change area, and it is determined whether the motion trajectory is a periodic identical motion, which can further determine whether the production process is proceeding normally and improve the accuracy of the early warning.
[0042] In another preferred embodiment of the present invention, during the monitoring process, if no machine stop command is collected, the shape or size of the motion trajectory changes, even if the motion trajectory is still a periodic and identical motion after the change, it is still determined to be an abnormal situation.
[0043] At this time, it is possible that the operator's operation has changed, or the production equipment has malfunctioned, causing abnormal equipment movement.
[0044] In another preferred embodiment of the present invention, during the monitoring process, if no machine stop instruction is collected, if the pixel area of the motion change area is reduced to less than a% of the original pixel area, a is a preset value, it is also determined to be an abnormal situation.
[0045] When the shape or size of the motion trajectory changes, or the pixel area of the motion change area decreases to less than a% of the original pixel area, it is also determined to be an abnormal situation. The value of a is first calculated as the ratio S1 / (S1+S2) of the equipment motion area S1 to the sum of the equipment motion area S1 and the personnel motion area S2. The value of a is 5%-10% larger than the value of the ratio. In this way, it is determined whether the personnel have left their posts without authorization or have not operated the equipment in accordance with the specifications, or whether the operator has an abnormal situation and his movement is restricted. This makes it possible to more comprehensively monitor abnormal situations in the production process and improve the reliability of early warning.
[0046] A safety production supervision and early warning system based on AI recognition, including:
[0047] A signal synchronization module, used to obtain monitoring instructions, including but not limited to electrical signals, action signals and voice signals, and including machine start instructions and machine stop instructions;
[0048] An image acquisition module is used to acquire the monitoring video in the production workshop when a machine start instruction is collected, split it to obtain a monitoring image sequence, and perform grayscale processing on the monitoring image sequence to generate a grayscale image sequence;
[0049] A change recognition module is used to recognize a preset dividing line in a grayscale image, wherein the preset dividing line is actually set in a production workshop, and a plurality of working area images are segmented from any grayscale image according to the preset dividing line, and a rectangular coordinate system is established with any fixed point of the working area image as the origin, and the working area image is recognized, and the working area image is divided into a background area and a change area according to the change of the grayscale value of the pixel point;
[0050] A motion recognition module is used to classify the pixels that are always in the changing area into the static changing area and the pixels that are intermittently in the changing area into the moving changing area according to the coordinates of each pixel in the changing area in the grayscale image sequence;
[0051] The trajectory determination module is used to obtain the coordinates of all pixel points in the motion change area in any working area image, regard the motion change area as equivalent to a plane entity with uniform density, calculate the center of mass coordinates of the motion change area according to the coordinates of all pixel points, generate a motion trajectory of the center of mass coordinates changing with time, and determine whether the motion trajectory is a periodic identical motion. If so, continue monitoring; if not, and no machine stop command is collected at this time, it is determined to be an abnormal situation, and the management personnel and safety personnel in the production workshop are prompted.
[0052] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.
Claims
1. A safety production supervision and early warning method based on AI recognition, characterized in that: The following steps are involved: Acquire monitoring instructions, wherein the monitoring instructions include a machine start instruction and a machine stop instruction; When a machine start instruction is collected, a monitoring video in the production workshop is obtained, split to obtain a monitoring image sequence, and grayscale processing is performed on the monitoring image sequence to generate a grayscale image sequence; Identify a preset dividing line in a grayscale image, the preset dividing line is actually set in the production workshop, segment a plurality of working area images from any grayscale image according to the preset dividing line, establish a rectangular coordinate system with any fixed point of the working area image as the origin, identify the working area image, and divide the working area image into a background area and a change area according to the change of the grayscale value of the pixel point; According to the coordinates of each pixel point in the change area in the grayscale image sequence, the pixel points that are always in the change area are divided into the static change area, and the pixel points that are intermittently in the change area are divided into the motion change area; For any motion change area in the image of the working area, the coordinates of all its pixel points are obtained, and the motion change area is equivalent to a planar entity with uniform density. The center of mass coordinates of the motion change area are calculated according to the coordinates of all its pixel points, and a motion trajectory of the center of mass coordinates changing with time is generated. It is determined whether the motion trajectory is a periodic identical motion. If so, monitoring continues; if not, and no machine stop command is collected at this time, it is determined to be an abnormal situation, and the management personnel and safety personnel in the production workshop are prompted.
2. According to the AI-based production safety supervision and early warning method of claim 1, it is characterized in that: When the monitoring instruction is an electrical signal or a voice signal, it is collected by an electrical signal collection module or a voice recognition module arranged in the production equipment; when the monitoring instruction is an action signal, it is collected by an image recognition module in the workshop monitoring camera.
3. According to the AI-based production safety supervision and early warning method of claim 2, it is characterized in that: The color or shape of the preset dividing line is different from the workshop background, and the color or shape of the preset dividing line is entered into the image recognition module.
4. According to the AI-based safety production supervision and early warning method of claim 1, it is characterized in that: The process of dividing the background area and the change area is as follows: Obtain the grayscale values (xi, yi) of all pixels of all grayscale images in the grayscale image sequence, where x and y represent the coordinates of any pixel in the grayscale image, i represents the time sequence of the grayscale image where the pixel is located in the entire grayscale image sequence, and i is a positive integer. By calculating the grayscale difference d of the pixels with the same coordinates between the grayscale images at adjacent times, if the grayscale difference d is greater than a preset illumination error threshold, the pixels with the same coordinates are the change area; if the grayscale difference d is less than the preset illumination error threshold, the pixels with the same coordinates are the background area.
5. According to the AI-based production safety supervision and early warning method of claim 4, it is characterized in that: The specific process of determining the static change area and the moving change area is as follows: Get the change area in the time series, collect the coordinates of the pixel points in the change area of each frame in turn, if there is a pixel point with any coordinates that is always in the change area, then the pixel point is determined to be a static change area; if there is a pixel point with any coordinates that belongs to both the change area and the background area in the time series, then the pixel point is a moving change area.
6. According to the AI-based production safety supervision and early warning method of claim 5, it is characterized in that: During the monitoring process, if the shape or size of the motion trajectory changes and no machine stop command is collected, it is still determined to be an abnormal situation even if the motion trajectory after the change is still the same periodic motion.
7. According to the AI-based safety production supervision and early warning method of claim 1, it is characterized in that: During the monitoring process, if no machine stop command is received, if the pixel area of the motion change area decreases to less than a% of the original pixel area, where a is a preset value, it is also determined to be an abnormal situation.
8. A safety production supervision and early warning system based on AI recognition, characterized in that: include: A signal synchronization module, used to obtain monitoring instructions, wherein the monitoring instructions include machine start instructions and machine stop instructions; An image acquisition module is used to acquire the monitoring video in the production workshop when a machine start instruction is collected, split it to obtain a monitoring image sequence, and perform grayscale processing on the monitoring image sequence to generate a grayscale image sequence; A change recognition module is used to recognize a preset dividing line in a grayscale image, wherein the preset dividing line is actually set in a production workshop, and a plurality of working area images are segmented from any grayscale image according to the preset dividing line, and a rectangular coordinate system is established with any fixed point of the working area image as the origin, and the working area image is recognized, and the working area image is divided into a background area and a change area according to the change of the grayscale value of the pixel point; A motion recognition module is used to classify the pixels that are always in the changing area into the static changing area and the pixels that are intermittently in the changing area into the moving changing area according to the coordinates of each pixel in the changing area in the grayscale image sequence; The trajectory determination module is used to obtain the coordinates of all pixel points in the motion change area in any working area image, regard the motion change area as equivalent to a plane entity with uniform density, calculate the center of mass coordinates of the motion change area according to the coordinates of all pixel points, generate a motion trajectory of the center of mass coordinates changing with time, and determine whether the motion trajectory is a periodic identical motion. If so, continue monitoring; if not, and no machine stop command is collected at this time, it is determined to be an abnormal situation, and the management personnel and safety personnel in the production workshop are prompted.
