Intelligent safety management system of digitized production line

Abstract

The application discloses an intelligent safety management system of a digital production line, which comprises a business layer, an algorithm layer and a data layer. The data layer acquires and stores real-time video data on the digital production line, and stores attention points and button points pre-marked in the real-time video data and corresponding to different action sequences of production line equipment action information. The business layer is connected with the production line equipment through a standardized interface and acquires real-time production line equipment action information, classifies the production line equipment action information into head action categories or hand action categories. The algorithm layer comprises an intelligent safety recognition algorithm, which is used to acquire human body images according to the real-time video data. According to the action classification, the application can realize different recognitions of the intelligent safety recognition algorithm, and is more targeted. The application realizes targeted compliance recognition and safety recognition, improves the recognition effect and the safety of the digital production line.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent management software of production line, and particularly relates to an intelligent safety management system of digital production line. BACKGROUND

[0002] The digital production line includes an automatic feeding and discharging and various manual workstations and a mobile AGV typical production line. The traditional SCADA system (Supervisory Control And Data Acquisition data acquisition and monitoring control system) can realize the collection of production data of the production line, but there are the following problems: the compliance rate and efficiency of the daily operation of the operators in the man-machine workstation and the manual workstation cannot be judged, and need to rely on post-mortem calculation; personnel action recognition, dangerous object recognition and preventive supervision cannot be realized by combining security and production management. In the identification of some operators, only part of the dangerous body action information can be identified, and targeted compliance recognition cannot be performed on the action information of the production line equipment, especially there is no effective means for attention recognition. That is, the existing technology has the problem of poor recognition efficiency and effect of compliance recognition of the digital production line. SUMMARY

[0003] Therefore, it is necessary to provide an intelligent safety management system of digital production line to solve the problem of poor recognition efficiency and effect of compliance recognition of the digital production line in the prior art.

[0004] To achieve the above purpose, the present application provides an intelligent safety management system of digital production line, which comprises a business layer, an algorithm layer and a data layer. The data layer acquires and stores real-time video data on the digital production line, and pre-marked attention points and button points in the real-time video data, and stores attention points and button points corresponding to different action sequences in the production line equipment action information. The business layer is connected with the production line equipment through a standardized interface and acquires real-time production line equipment action information, classifies the production line equipment action information into head action category or hand action category. The algorithm layer comprises an intelligent safety recognition algorithm. The intelligent safety recognition algorithm is used to acquire a human body image according to the real-time video data, recognize a trunk node, an arm node and a head node according to the human body image, calculate a head direction according to the head node when classified as the head action category, judge whether the head direction is within a preset range of the attention point corresponding to the action sequence in the production line equipment action information, and perform an abnormal alarm if it is not within the preset range. When classified as the hand action category, the trunk node and the arm node are calculated, and the button point corresponding to the action sequence in the production line equipment action information is calculated, and the arm angle information is calculated. If the arm angle information exceeds the preset range, an abnormal alarm is performed.

[0005] Further, the intelligent safety identification algorithm is further used for identifying a current action sequence according to the production line equipment action information, judging whether the action sequence meets a preset standard action sequence, and performing an abnormal alarm if the standard action sequence is not met.

[0006] Further, the intelligent safety identification algorithm includes a YOLOv5n detection algorithm and a Lite-HRNet detection algorithm, a SimAM module is embedded in a backbone of the YOLOv5n detection algorithm, a human body image is identified by using the embedded YOLOv5n detection algorithm, a Conv2D in the Lite-HRNet is replaced by a GSConv, a FPN upsampling layer of the Lite-HRNet detection algorithm is replaced by a CARAFE, and the Lite-HRNet detection algorithm is used to identify a torso node, an arm node and a head node, and return names and coordinates of the nodes.

[0007] Further, the data layer pre-stores different dangerous areas of different nodes, and the intelligent safety identification algorithm is further used for judging whether the identified torso node, arm node and head node are in the dangerous areas respectively, and performing an abnormal alarm if the nodes are in the dangerous areas.

[0008] Further, the intelligent safety identification algorithm is further used for calculating a horizontal angle of the torso according to the identified torso node, and performing an abnormal alarm if the angle is greater than a preset value.

[0009] Further, the abnormal alarm is divided into different levels, including a system record prompt, an audible and light alarm, different alarm modules are driven according to the abnormal alarm level to perform an alarm, and the alarm modules include voice broadcast, red light warning, display screen prompt and the like.

[0010] Further, when the abnormal alarm occurs, the business layer records product ID information of a current product and establishes a corresponding relationship between the abnormal alarm and the product ID information in the data layer.

[0011] Further, a detection module is further included, which is used for acquiring product ID information of the data layer and detecting a product corresponding to the product ID information.

[0012] Further, the production line equipment is a lithium battery processing production line equipment.

[0013] Further, a web interaction layer is further included, which acquires login information of a user, displays a video screen, and is further used for acquiring a marking information update operation of the user to re-mark a focus point and a button point.

[0014] Different from the prior art, the technical scheme can identify the production line action through the production line equipment action information, and then perform different action classification and action corresponding attention points and button points, and can realize different identification of the intelligent safety identification algorithm according to the action classification, and is more targeted. The intelligent safety identification algorithm identifies the human body image and the torso node, the arm node and the head node, and through the nodes, the head direction or the arm angle can be calculated in a targeted manner, so as to realize the identification of the operator's attention and the operation safety, and finally realize the targeted compliance identification and safety identification, improve the identification effect, and improve the safety of the digital production line. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 It is a system architecture diagram of the present application.

[0016] Figure 2 It is an algorithm architecture diagram of the intelligent safety identification algorithm of the present application.

[0017] Figure 3 It is a system flow logic diagram of the present application.

[0018] Figure 4 It is a video image identification process diagram of the intelligent safety identification algorithm of the present application. DETAILED DESCRIPTION

[0019] In order to specifically describe the technical content, structural features, purposes and effects of the technical scheme, the following will be described in detail in combination with specific embodiments and the accompanying drawings.

[0020] Please refer to Figures 1 to 4 The present application provides an intelligent safety management system for digital production line, which comprises a business layer, an algorithm layer and a data layer. The data layer acquires and stores real-time video data on the digital production line and attention points and button points marked in advance in the real-time video data. It should be noted that the real-time video data is collected by a fixed-angle camera, which is generally fixed beside the production line (referred to as the production line) equipment. The attention point here is a point that needs to be viewed by the operator, such as Figure 4 as shown in Figure 4 the upper middle display screen of Figure 4In the center are multiple buttons of different colors, labeled as button points. The system also stores focus points and button points corresponding to different action sequences in the production line equipment's action information. Here, "correspondence" means that different action steps on the production line correspond to different focus points or button points. For example, in visual inspection, the focus point is a green dot on the display screen, and in the action step of pressing the red button, the corresponding button point is a red button point. The business layer connects to the production line equipment through a standardized interface to obtain real-time production line equipment action information. Here, "standardized interface" refers to a unified interface, using a specific standard interface (such as a CAN bus) for connection. The production line equipment action information refers to the action information triggered by the production line equipment. For example, if the red button is triggered, action information for the red button being triggered will be generated. If visual inspection displays an image on the screen, action information for visual inspection will be generated. The production line equipment action information is categorized into head action categories or hand action categories. This categorization can be pre-defined; for example, pressing a button or tightening a screw is a hand action category, while visual inspection is a head action category. This category is subsequently used for recognition and differentiation by the algorithm layer.

[0021] The algorithm layer includes an intelligent security recognition algorithm, which can be pre-configured into the system using a neural network recognition algorithm. This intelligent security recognition algorithm is used to acquire human images based on real-time video data, such as... Figure 4 As shown, the yellow boxes correspond to the recognized human body images. Then, based on the human body image, trunk nodes, arm nodes, and head nodes are identified, such as... Figure 4 The blue dots in the human body image above indicate this. When categorized as a head movement, the head direction is calculated based on the head nodes. Specifically, the forward direction the head points can be calculated by connecting multiple head nodes. Then, it is determined whether the head direction falls within the preset range of the focus points corresponding to the movement sequence in the production line equipment's movement information. For example... Figure 4 In the process, when the production line equipment is in a visual inspection sequence, it checks whether the head direction is within the preset range of the green dots on the display screen. If it is not within the preset range, an abnormal alarm is triggered. This preset range can be the vertical distance from the blue focus point to the direction in front of the head. If it is within the preset range, a visual inspection is considered to have been performed; otherwise, it is considered not to have been performed, and an abnormal alarm is triggered. When categorized as a hand movement, the arm angle information is calculated based on the torso and arm nodes, as well as the button points corresponding to the action sequence in the production line equipment's action information. If the arm angle information exceeds the preset range, an abnormal alarm is triggered. The arm angle includes the upper arm angle and the lower arm angle. If it is not within a reasonable range, it indicates that the arm movement is not standardized and may cause safety issues.

[0022] The application can identify the action of the production line through the production line equipment action information, and then perform different action classification and action corresponding attention points and button points, and different recognitions of the intelligent safety recognition algorithm can be realized according to the action classification, which is more targeted. The intelligent safety recognition algorithm is used for recognizing the human body image and the torso node, the arm node and the head node, and through these nodes, the head direction or the arm angle can be calculated in a targeted manner, so as to realize the recognition of the attention of the operator and the operation safety, and finally realize the targeted compliance recognition and safety recognition, improve the recognition effect, and improve the safety of the digital production line.

[0023] Further, the intelligent safety recognition algorithm is also used for identifying the current action sequence according to the production line equipment action information, judging whether the action sequence meets the preset standard action sequence, and performing abnormal alarm if the standard action sequence is not met. The system of the application is also used for identifying the action sequence and comparing it with the preset standard action sequence, and if the sequence does not meet the standard, an abnormal alarm is triggered. By combining the current state of the equipment with the human body action data, it is judged whether the operation process is compliant, and if there is a step disorder, such as incorrect button sequence or pressing multiple buttons, which violates the operation specification, the system can timely discover. The embodiment solves the problem that the operation process in the production line is not standardized and is difficult to be discovered in real time, and improves the monitoring ability of the process integrity and the operation sequence.

[0024] In order to improve the detection effect, as shown in Figure 2 The intelligent safety recognition algorithm includes a YOLOv5n detection algorithm and a Lite-HRNet detection algorithm, a SimAM module is embedded in the backbone of the YOLOv5n detection algorithm, the embedded YOLOv5n detection algorithm is used to realize human body image recognition, and then the Conv2D in the LiteHRNet is replaced by the GSConv, and the FPN upsampling layer of the Lite-HRNet detection algorithm is replaced by the CARAFE, and the Lite-HRNet detection algorithm is used to realize the identification of the torso node, the arm node and the head node, and the names and coordinates of each node are returned. The application adopts the improved YOLOv5n and Lite-HRNet detection algorithm. The SimAM module embedded in the YOLOv5n is used for human body detection, and the Lite-HRNet replaces part of the modules to enhance the key point detection accuracy, which is used to identify the torso, head and arm nodes and their coordinates. In the working process, the YOLOv5n is responsible for detecting the whole human body frame, and the Lite-HRNet performs fine-grained identification on the key parts in the frame. Through the structural optimization of the neural network, the detection speed and accuracy can be improved, and the real-time performance is improved.

[0025] In the implementation of the application, reliability and real-time performance verification is also needed, and the logic diagram of the verification is as follows Figure 3As shown, the accuracy requirement and timeliness requirement of identification are required to be further optimized when the requirements are not met. The front-end data display can be realized by jsp, css and html to realize the display of the webpage.

[0026] In some embodiments, the data layer pre-stores different dangerous areas of different nodes, and the intelligent safety identification algorithm is further used to judge whether the trunk node, the arm node and the head node are in the dangerous area respectively according to the identified nodes, and if yes, an abnormal alarm is performed. Figure 4 As shown, the right red box is the dangerous area of the head node and the trunk node, if the head enters the area, danger may occur, and the system performs an alarm. The data layer of the present application pre-stores the dangerous area information corresponding to different nodes, and the intelligent safety identification algorithm can judge whether the operator enters the dangerous area after identifying the node coordinates, and trigger the corresponding abnormal alarm. The safety hidden danger can be automatically judged, and the safety is improved.

[0027] Further, the intelligent safety identification algorithm is further used to calculate the angle between the trunk and the horizontal according to the identified trunk node, and if the angle is greater than a preset value, an abnormal alarm is performed. The angle between the trunk node and the horizontal line is calculated, and if it exceeds the set range, an abnormal alarm is triggered. The present application can analyze the posture angle of the personnel, such as whether the trunk inclination angle is abnormal, which helps to detect non-standard postures such as excessive bending, low head operation, fainting, etc. Thus the safety of the digital production line is improved, and the possibility of injury caused by abnormal posture is reduced.

[0028] In some embodiments, the abnormal alarm is divided into different levels, including system record prompt, sound and light alarm, different alarm modules are driven according to the abnormal alarm level to perform alarm, and the alarm modules include voice broadcast, red light warning, display screen prompt, etc. The present embodiment realizes the grading strategy of the alarm mechanism. The system can trigger different levels of prompts such as recording, voice broadcast, red light warning, display screen reminder according to the abnormal degree. In the working process, the intelligent safety identification algorithm outputs the alarm level, and the alarm module selects the appropriate output form to feedback to the operator. Through diversified prompting means, the effectiveness of the warning is improved.

[0029] Further, when the abnormal alarm occurs, the business layer records the product ID information of the current production and establishes the corresponding relationship between the abnormal alarm and the product ID information in the data layer. When the abnormal alarm occurs, the business layer should record the product ID corresponding to the current production line, and establish the association between the product and the abnormality in the data layer. In the running process of the present application, each alarm not only triggers a prompt, but also automatically traces to a specific product, realizing the binding of the product and the abnormal data. Thus the association management of the product and the quality accountability is realized, providing a basis for subsequent traceability.

[0030] Further, a detection module is further included for acquiring product ID information of the data layer and detecting a product corresponding to the product ID information. The application adds the function of the detection module: the module can acquire product ID and detect the state of the corresponding product based on the product ID. In normal cases, the production line adopts sampling inspection, and when an alarm occurs, the product at the time of the alarm is detected in real time, which can further improve the detection effect and avoid the occurrence of abnormal products due to non-standard operation. After identifying the abnormality, the detection module performs quality detection on the corresponding product. The problem of abnormal behavior of the digital production line not being able to be linked to the quality inspection process is solved, the efficiency of abnormal processing is improved through the function, and the accuracy of abnormal product rejection is further ensured.

[0031] In some embodiments, the production line equipment is lithium battery processing production line equipment. The application can be applied to the high-risk and high-precision process requirement lithium battery industry, thereby improving the product qualification rate of lithium battery production and avoiding safety problems of the production line.

[0032] Further, a web interaction layer is further included, which acquires login information of a user and displays a video screen, and the web interaction layer is further used for acquiring a marking information updating operation of the user to re-mark a focus point and a button point. Through the web interaction layer, the application allows the user to log in, view the video and re-mark the focus point and the button point through the Web page. The interaction layer supports the administrator to perform system setting after system deployment, and the user can adjust the marking point according to the change of the camera or the change of the production line equipment, thereby improving the adaptability. The problem of fixed pre-annotation and difficult flexible adjustment is solved.

[0033] It should be noted that although the above embodiments have been described in the present text, the patent protection scope of the application is not limited thereby. Therefore, based on the innovative idea of the application, changes and modifications of the embodiments described in the present text, or equivalent structures or equivalent process transformations made by using the content of the application specification and drawings, direct or indirect application of the above technical solutions to other related technical fields, are all included in the patent protection scope of the application.

Claims

1. An intelligent safety management system for digitizing a production line, characterized by: The system comprises a business layer, an algorithm layer and a data layer, the data layer acquires and stores real-time video data on a digitized production line, and pre-labeled attention points and button points in the real-time video data, and stores attention points and button points corresponding to different action sequences in production line equipment action information, and the corresponding attention point in visual inspection is a green point on a display screen; the business layer is connected with the production line equipment through a standardized interface and acquires real-time production line equipment action information, classifies the production line equipment action information into a head action category or a hand action category; the algorithm layer comprises an intelligent safety recognition algorithm, the intelligent safety recognition algorithm is used for acquiring a human body image according to real-time video data, and identifying a torso node, an arm node and a head node according to the human body image; when classified into the head action category, a head direction is calculated according to the head node, a front direction pointed by the head is calculated according to a plurality of head node connecting lines, and it is judged whether the head direction is within a preset range of the attention point corresponding to the action sequence in the production line equipment action information, and if not, an abnormal alarm is given, when the production line equipment is in a visual inspection action sequence, it is judged whether the head direction is within a preset range of the green point on the display screen, and if not, an abnormal alarm is given; when classified into the hand action category, a hand arm angle information is calculated according to the torso node and the arm node and the button point corresponding to the action sequence in the production line equipment action information, and if the hand arm angle information exceeds a preset range, an abnormal alarm is given. The intelligent safety recognition algorithm comprises a YOLOv5n detection algorithm and a Lite-HRNet detection algorithm, a SimAM module is embedded in a backbone of the YOLOv5n detection algorithm, the human body image is identified by using the embedded YOLOv5n detection algorithm, then Conv2D in LiteHRNet is replaced by GSConv, and a FPN upsampling layer of the Lite-HRNet detection algorithm is replaced by CARAFE, and the torso node, the arm node and the head node are identified by using the Lite-HRNet detection algorithm, and the names and coordinates of the nodes are returned.

2. The intelligent safety management system for digitized production line as claimed in claim 1 wherein: The intelligent safety recognition algorithm is also used for identifying a current action sequence according to the production line equipment action information, judging whether the action sequence meets a preset standard action sequence, and giving an abnormal alarm if the standard action sequence is not met.

3. The intelligent safety management system for digitized production line as claimed in claim 1 wherein: The data layer pre-stores different dangerous areas of different nodes, and the intelligent safety recognition algorithm is also used for judging whether the identified torso node, arm node and head node are within the dangerous areas respectively, and giving an abnormal alarm if the nodes are within the dangerous areas.

4. The intelligent safety management system for digitized production line of claim 1, wherein: The intelligent safety recognition algorithm is also used for calculating a torso and horizontal angle according to the identified torso node, and giving an abnormal alarm if the angle is greater than a preset value.

5. The intelligent safety management system for digitized production line as claimed in claim 1 wherein: The abnormal alarm is divided into different levels, including system record prompts, sound and light alarms, different alarm modules are driven according to the abnormal alarm levels to give alarms, and the alarm modules include voice broadcasting, red light warnings and display screen prompts.

6. The intelligent safety management system for digitized production line of claim 1, wherein: When an abnormal alarm occurs, the business layer records current product ID information produced and establishes a corresponding relationship between the abnormal alarm and the product ID information in the data layer.

7. The intelligent safety management system for digitized production line as claimed in claim 6 wherein: The detection module is further configured to acquire product ID information of the data layer and detect a product corresponding to the product ID information.

8. The intelligent safety management system for digitized production line of claim 1, wherein: The production line equipment is lithium battery processing production line equipment.

9. The intelligent safety management system for digitized production line as claimed in claim 1 wherein: The web interaction layer is further configured to acquire login information of a user, display a video screen, acquire a marking information update operation of the user, and re-mark a focus point and a button point.

Images