Security monitoring method and system for production process of barrel type equipment

Abstract

The invention discloses a safety monitoring method and system for a cylinder equipment production process, and relates to the technical field of industrial safety monitoring, and the method comprises the steps: collecting a video stream of a cylinder equipment production region in real time, and uploading the video stream; automatically identifying which flow section of the production process the current operation content belongs to according to the video stream information, if no identification result exists, returning to the step 1, and continuously and circularly monitoring; if the identification result exists, calling a corresponding detection algorithm to identify the security risk existing in the production of the flow section according to the identified flow section; outputting a safety monitoring result from the obtained detection result according to a set risk judgment condition; if the monitoring result is that no safety risk exists, returning to the step 1; and if the monitoring result is that the security risk exists, triggering a security alarm. According to the method, different scenes in the whole production process are classified and identified according to the existing risk characteristics, so that safety accidents can be effectively prevented, the false identification rate is reduced, and the overall production safety level and the production efficiency are further improved.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial safety monitoring, and in particular to a method and system for monitoring the safety of a production process of a cylinder equipment. Background Art

[0002] Cylinder equipment is widely used in the industrial field. For example, the hot air leaf moistening machine cylinder can be used to increase the temperature and humidity of tobacco leaves in the tobacco processing industry; the drying cylinder of the tobacco drying machine can be used to evenly dry the tobacco.

[0003] At the same time, the production process of cylinder equipment is also relatively complicated. Specifically, the steel plate needs to be rolled into a roll by a plate rolling machine. Then several rolls are tightly spliced ​​into a large cylinder through welding. Finally, the required parts are welded inside the cylinder. There are many potential safety risks in this process, so some necessary protective measures need to be taken. For example: since the hoisted objects may slip during hoisting operations, people cannot stand under the hoisted objects; when operating the plate rolling machine, the operator must wear goggles to avoid injury from splashes; when welding externally, welding work clothes, protective gloves and protective masks must be strictly worn; in addition, it is forbidden to place debris on the surface of the roll to prevent slipping and causing injuries to people below; safety belts must be strictly worn when welding at heights; when welding internally, since the working space is relatively closed, fans must be equipped for auxiliary ventilation, and welding work clothes, protective gloves and protective masks must be strictly worn.

[0004] At present, it is difficult to achieve real-time and accurate monitoring of the entire processing process through manual monitoring, and it is easy to have missed detection and false detection. There is no safety monitoring method and system dedicated to the production process of tube equipment. At the same time, most of the current visual inspection methods cannot adapt to the needs of multiple scenarios and multiple changes. They can only target a single scenario, and the operation is complex and inefficient. For example, patent CN202310818645.5 proposes a safe welding system and method based on Yolov5, including: a robotic arm and a control module, the robotic arm is equipped with a welding gun, the control module is connected to the control end of the robotic arm to control the robotic arm to weld the workpiece to be welded; a camera input module, used to obtain video source data of the work site; a safety helmet detection module, which is equipped with a first target detection model based on Yolov5, used to identify the workers in the video source data, and to identify whether the workers are wearing safety helmets, and output safety helmet detection results; a workpiece quality inspection module, which is equipped with a second target detection model based on Yolov5, used to perform quality inspection on the workpieces in the video source data, and output welding quality inspection results; a display module, used to visually display the safety helmet detection results and the welding quality inspection results.

[0005] Therefore, there is an urgent need for a method that can automatically identify safety risks in the entire production process of barrel equipment to improve production safety and efficiency. Summary of the invention

[0006] The purpose of the present invention is to provide a safety monitoring method and system specifically for the production process of cylinder equipment in response to the above-mentioned problems. By real-time monitoring of the safety behavior of the entire cylinder equipment production process, it is possible to effectively warn and prevent the occurrence of safety accidents, thereby improving the overall production safety level and production efficiency; and in monitoring and identification, segmented identification is performed, first judging the process segment, and then further identifying and monitoring the different safety risks of different process segments, effectively reducing the complexity of identification, and more quickly and accurately identifying safety risks in the production process.

[0007] The technical solution of the present invention is as follows: The present invention discloses a method for safety monitoring of a production process of a cylinder equipment, comprising the following steps: Step 1: Collect the video stream of the production area of ​​the drum equipment in real time and upload it; Step 2: Automatically identify which process segment of the production process the current operation content belongs to based on the video stream information. The process segments include the hoisting segment, the roll forming segment, the external welding segment, and the internal welding segment. If there is no recognition result, return to step 1 and continue the cycle monitoring. If there is a recognition result, call the corresponding detection algorithm based on the recognized process segment to identify the safety risks in the production of the process segment. Step 3: Output the security monitoring result of the identification result obtained in step 2 according to the set risk judgment conditions; if the monitoring result is no security risk, return to step 1 and continue the loop monitoring; if the detection monitoring shows that there is a security risk, trigger the security alarm, and then return to step 1 and continue the loop monitoring.

[0008] Furthermore, the identification feature of the hoisting section in step two is: the crane is performing hoisting operations and there are objects hoisted; the identification feature of the drum forming section is: the plate rolling machine is in operation and the steel plate is being rolled; the identification feature of the external welding section is: the staff is performing welding operations on the outside of the drum; and the identification feature of the internal welding section is: the staff is performing welding operations inside the drum.

[0009] Furthermore, the specific implementation method of step 2 includes: S21: Extract the videos collected in step 1, including the videos of hoisting operation, plate rolling machine plate rolling, external welding of the roll, and internal welding of the roll; divide the video into a segment every 10 frames, label each segment with a category, and construct an action recognition dataset; S22: extracting target images including a hook, a reel, a human body, goggles, a protective mask, protective gloves, welding overalls, a safety belt, sundries and a fan from the video collected in step 1; annotating the targets in the target images, generating an annotation file including target position and category information, and constructing a target detection dataset; S23: using the action recognition algorithm and the target detection algorithm to train the action recognition dataset and the target detection dataset respectively; S24: Use the trained action recognition model and target detection model to identify which process segment the current operation content belongs to and the security risks existing in the process segment.

[0010] Furthermore, in step 2, if the current process section is the hoisting section, detecting whether there is a person standing under the hoisted object specifically includes: First, identify the hook and the person, and obtain the coordinate information of the human body detection frame as [ X min-p , Y min-p , X max-p , Y max-p ], the coordinate information of the hook detection frame is [ X min-h , Y min-h , X max-h , Y maxh ]; Secondly, calculate the horizontal Euclidean distance between the center points of the hook detection frame and the human body detection frame.

[0011] Finally, if the horizontal Euclidean distance d Less than the specified threshold d th , it means that there is a person standing under the suspended object; if the horizontal Euclidean distance d Greater than a specified threshold d th , it means there is no one standing under the suspended object.

[0012] Furthermore, in step 2, if the current process segment is the roll forming segment, check whether the plate rolling machine operator wears goggles; if the current process segment is the external welding segment, check whether there are debris on the roll, whether the welders wear welding work clothes, protective gloves and protective masks, and whether they wear safety belts for high-altitude operations; if it is detected that there are debris on the roll or the welders are not wearing protective equipment, directly output that there is a safety risk; if there is no debris on the roll and the welders are wearing protective equipment, further determine whether there is high-altitude operation, and if there is no high-altitude operation, output that there is no safety risk; if there is high-altitude operation, further determine whether a safety belt is worn, and if not, output that there is a safety risk.

[0013] Further, the judgment and identification method of high-altitude operations: S31: Obtain the coordinate information of the human body detection frame as [ X min-p , Y min-p , X max-p , Y max-p ], the coordinate information of the roll detection frame is [ X min-r , Y min-r , X max-r , Y max-r ]; S32: Calculate the height difference between the human body and the center point of the reel: ; S33: Calculate the intersection and union ratio of the human body and the roll: ,in, ; S P and S R are the areas of the human body detection frame and the roll detection frame respectively; S32: Based on the intersection and union ratio of the human body detection frame and the roll detection frame IOU The height difference between the center point h Determine whether a person is working at height; the conditions for determining whether a person is working at height on the drum are: h>h th and 0 <IOU<IOU th , h th is the height difference threshold between the human body and the center point of the reel. IOU th The threshold of the intersection ratio between the human body and the reel is set.

[0014] Furthermore, in step 2, if the current process segment is the internal welding segment, it is detected whether the welder is wearing welding overalls, protective gloves and protective masks, and whether a fan is placed outside the reel to assist ventilation; the wearing of welding overalls, protective gloves and protective masks, and the placement of the fan are detected in parallel, and as long as none of them is detected, a safety alarm is triggered.

[0015] Furthermore, in step three, the risk assessment conditions include: people standing under the suspended objects; the plate rolling machine operator not wearing goggles; the welder not wearing welding work clothes, protective gloves and protective masks; the welder not wearing a safety belt when working at height; the welder working inside the cylinder without fan-assisted ventilation.

[0016] The present invention also discloses a safety monitoring system for the production process of cylinder equipment, including a video acquisition module, a safety monitoring module and an alarm module; The video acquisition module is used to acquire the video stream of the production area in real time; The safety monitoring module is used to first identify which process segment of the production process the current operation content belongs to, and then use corresponding methods according to different process segments to detect safety risks in the production of the process segment; The alarm module generates an alarm based on the type of event triggered by the security risk.

[0017] Further, the safety monitoring module includes a classification submodule and a detection submodule; The classification submodule uses a video understanding algorithm to automatically identify the process segment of the current operation, the hoisting segment, the roll forming segment, the external welding segment or the internal welding segment; The detection submodule calls corresponding detection algorithms for each process segment to identify the security risks of different process segments.

[0018] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: 1. The present invention realizes real-time and accurate monitoring of the whole process of cylinder equipment production, and is dedicated to cylinder equipment production, solving the problem that there is no safety monitoring system dedicated to the cylinder equipment production process.

[0019] 2. The present invention can automatically monitor and identify safety risks in the production process, effectively warn and prevent the occurrence of safety accidents, and solve the problem that manual monitoring methods are difficult to achieve real-time and accurate monitoring of the entire processing process, and are prone to missed detections and false detections.

[0020] 3. The present invention improves the efficiency of video surveillance identification. In video identification processing, different from the existing method of judging and identifying all behaviors and features in the production process, based on the characteristics of the production process of cylinder equipment, the process segment is first judged, and then the specific process segment is judged. The whole process is carried out in sections, and the analysis and judgment can be focused, which simplifies the judgment and identification methods, is more targeted, and has higher processing efficiency.

[0021] 4. The present invention designs a detection algorithm specifically for the production process of cylinder equipment, and the detection efficiency is significantly improved.

[0022] 5. The present invention can effectively prevent the occurrence of safety accidents and reduce the misidentification rate by classifying and identifying different scenarios in the entire production process according to their risk characteristics, thereby improving the overall production safety level and production efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The present invention will now be described by way of example with reference to the accompanying drawings, in which: Figure 1 It is a flow chart of a method for safety monitoring of a production process of a cylinder type equipment according to the present invention.

[0024] Figure 2 Schematic diagram of the steps for identifying safety risks of external welding sections in an embodiment of the present invention.

[0025] Figure 3 The present invention is a schematic flow chart of a safety monitoring system for a production process of a cylinder type equipment. DETAILED DESCRIPTION

[0026] All features disclosed in this specification, or steps in all methods or processes disclosed, except mutually exclusive features and / or steps, can be combined in any manner.

[0027] Any feature disclosed in this specification (including any additional claims and abstract), unless otherwise stated, may be replaced by other alternative features that are equivalent or have similar purposes. That is, unless otherwise stated, each feature is only an example of a series of equivalent or similar features.

[0028] The features and performance of the present invention are further described in detail below in conjunction with the embodiments.

[0029] The present invention provides a method for safety monitoring of a production process of a cylinder equipment, comprising the following steps: Step 1: Collect the video stream of the production area of ​​the drum equipment in real time and upload it; Step 2: Automatically identify which process segment of the production process the current operation content belongs to based on the video stream information. The process segments include the hoisting segment, the roll forming segment, the external welding segment, and the internal welding segment. If there is no recognition result, return to step 1 and continue the cycle monitoring. If there is a recognition result, call the corresponding detection algorithm based on the recognized process segment to identify the safety risks in the production of the process segment. Step 3: Output the security monitoring result of the identification result obtained in step 2 according to the set risk judgment conditions; if the monitoring result is no security risk, return to step 1 and continue the loop monitoring; if the monitoring result is a security risk, trigger a security alarm, and then return to step 1 and continue the loop monitoring.

[0030] like Figure 1 As shown, in one embodiment, a method for safety monitoring of the production process of a cylinder equipment is disclosed, and the specific steps include: Step 1: The video acquisition module collects the video stream of the barrel equipment production area in real time and inputs it into the safety monitoring module.

[0031] Step 2: The classification submodule in the safety monitoring module automatically identifies the process segments of the current operation content according to the video stream information, including the lifting segment, roll forming segment, external welding segment, and internal welding segment; if there is no classification result, it means that there is no operation in the production area at present, and returns to step 1 for continuous loop monitoring.

[0032] Step 3: The detection submodule calls the corresponding detection algorithm for the current process segment to identify security risks.

[0033] If the current process section is the hoisting section, check whether there is a person standing under the hoisted object; If it is the roll forming section, check whether the plate rolling machine operator is wearing goggles; If it is an external welding section, check whether there are any debris on the reel; whether the welders are wearing welding overalls, protective gloves and protective masks; and whether they are wearing safety belts when working at heights; If it is an internal welding section, check whether the welders are wearing welding overalls, protective gloves and protective masks, and whether there is a fan outside the reel to assist ventilation.

[0034] Step 4: Analyze the identification results output in step 3, and output the safety monitoring results according to the set risk judgment conditions; if the result is no safety risk, return to step 1 and continue the loop monitoring; if there is a safety risk, trigger a safety alarm, and then return to step 1 and continue the loop monitoring.

[0035] The risk assessment conditions include: people standing under the hanging objects; the plate rolling machine operator not wearing goggles; the welder not wearing welding work clothes, protective gloves and protective masks; the welder not wearing a safety belt when working at height; the welder working inside the cylinder without fan-assisted ventilation.

[0036] Step 5: The alarm module broadcasts the event in the production area in the form of a voice broadcast according to the type of event triggered; at the same time, it sends a notification to the workshop management personnel so that they can take corresponding measures in time.

[0037] The specific implementation methods of step 2 include: Data collection: Collect operation videos of lifting operations, plate rolling machine rolling, external welding of the roll, and internal welding of the roll. Data processing: Divide the video into a segment every 10 frames, label each segment with a category, and construct an action recognition dataset. The category labels of the video segments are lifting segment, roll forming segment, external welding segment, and internal welding segment. The characteristics of the lifting segment are that the crane is performing lifting operations and there are objects hanging. The characteristics of the roll forming segment are that the plate rolling machine is in operation and the steel plate is being rolled. The characteristics of the external welding segment are that the staff is performing welding operations on the outside of the roll. The characteristics of the internal welding segment are that the staff is performing welding operations on the inside of the roll. Model training: The TSM-ResNet18 action recognition algorithm is used to train the action recognition dataset, and the weight model trained by the public dataset Kinetics is used as the pre-training model. Process segment recognition: The trained action recognition model is used to analyze the video stream information collected in step one in real time to identify the process segment of the current operation content. The specific implementation method of step three includes: Data collection: Extract images containing hooks, reels, human bodies, goggles, protective masks, protective gloves, welding overalls, safety belts, debris, fans and other targets in the video. Data processing: Use image annotation tools to annotate the targets in the image, generate annotation files containing target location and category information, and construct target detection datasets. Model training: The yolov8n target detection algorithm is used to train the target detection dataset, and the weight model trained by the public dataset coco128 is used as a pre-training model to obtain four algorithm models, including the hoisting segment algorithm, the reel forming segment algorithm, the external welding segment algorithm, and the internal welding segment algorithm. Among them, the hoisting segment algorithm trains targets such as human bodies, reels, and hooks; the reel forming segment algorithm trains targets such as goggles and human bodies; the external welding segment algorithm trains targets such as human bodies, protective masks, protective gloves, welding overalls, safety belts, and debris. The internal welding segment algorithm trains targets such as human bodies, protective masks, protective gloves, welding overalls, fans and other targets. Safety risk identification: According to the process segment identified in step 2, enable the corresponding detection algorithm to identify the safety risks existing in the production of the process segment. like Figure 2As shown, in the external welding section in step 3: if it is detected that the protective equipment (welding overalls, protective gloves and protective masks) is not worn or there are debris on the reel, it is directly output that there is a safety risk; if there is no debris on the reel and the welder is wearing protective equipment, it is further determined whether there is any high-altitude work. If there is no high-altitude work, it is output that there is no safety risk; if there is any high-altitude work, it is further determined whether a safety belt is worn. If not, it is output that there is a safety risk.

[0038] In step 3, the internal welding section: welding overalls, protective gloves and protective masks, and fans are tested in parallel. If any one of them is not detected, a safety alarm is triggered.

[0039] The detection method of the person standing under the hanging object in step 3 is: first identify the hook and the person, and obtain the coordinate information of the human body detection frame as [ X min-p , Y min-p , X max-p , Y max-p ], the coordinate information of the hook detection frame is [ X min-h , Y min-h , X max-h , Y maxh ]; Secondly, calculate the horizontal Euclidean distance between the center points of the hook detection frame and the human body detection frame , if the horizontal Euclidean distance d Less than the specified threshold d th , it means that there is a person standing under the hanging object; if the horizontal Euclidean distance d Greater than a specified threshold d th , it means there is no one standing under the suspended object.

[0040] The method for identifying high-altitude work in step 3 is: based on the intersection and union ratio of the human body detection frame and the roll detection frame IOU The height difference between the center point and the human body is used to determine whether the person is working at height; the coordinate information of the human body detection frame is [ X min-p , Y min-p , X max-p , Y max-p ], the coordinate information of the roll detection frame is [ X min-r , Y min-r , X max-r , Y max-r ]; Height difference between human body and center point of reel: ; The intersection ratio of the human body and the reel: ; in:

[0041] and They are the areas of the human body detection frame and the roll detection frame respectively.

[0042] The judgment conditions for people to work at height on the drum are: h > h th and 0 <IOU<IOU th It means that the human body is above the reel and there is a certain overlap between the reels, ensuring that the human body is indeed near the reel and avoiding misjudging people in the distance as climbing work.

[0043] In the embodiment of the present invention, d th =200Px, h th =160Px, IOU th =0.3, the threshold can be adjusted according to the actual engineering environment d th , h th and IOU th .

[0044] like Figure 3 As shown, the present invention also discloses a safety monitoring system for the production process of drum equipment, including a video acquisition module, a safety monitoring module and an alarm module; the safety monitoring module includes a classification submodule and a detection submodule. The video acquisition module acquires the video stream of the production area in real time, the classification submodule in the safety monitoring module uses a video understanding algorithm to automatically identify the process segments described in the current operation (hoisting segment, roll forming segment, external welding segment, internal welding segment), and the detection submodule calls the corresponding detection algorithm for each process segment to identify the safety risks of different process segments; the alarm module makes corresponding voice broadcasts according to the event type triggered by the safety risk, and sends notifications to the workshop management personnel at the same time. By classifying and identifying different scenarios in the entire production process according to their existing risk characteristics, it can effectively warn and prevent the occurrence of safety accidents, and reduce the misidentification rate, thereby improving the overall production safety level and production efficiency.

[0045] The above-mentioned embodiments only express the specific implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the protection scope of the present application. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the technical solution concept of the present application, and these all belong to the protection scope of the present application.

Claims

1. A method for safety monitoring of the production process of cylinder equipment, characterized in that: The following steps are involved: Step 1: Collect the video stream of the production area of ​​the drum equipment in real time and upload it; Step 2: Automatically identify which process segment of the production process the current operation content belongs to based on the video stream information. The process segments include the hoisting segment, the roll forming segment, the external welding segment, and the internal welding segment. If there is no recognition result, return to step 1 and continue the cycle monitoring. If there is a recognition result, call the corresponding detection algorithm based on the recognized process segment to identify the safety risks in the production of the process segment. Step 3: Output the safety monitoring result based on the identification result obtained in step 2 according to the set risk judgment conditions; If the monitoring result is that there is no safety risk, then return to step one and continue the loop monitoring; if the monitoring result is that there is a safety risk, then trigger a safety alarm, then return to step one and continue the loop monitoring.

2. The method for safety monitoring of the production process of cylinder equipment according to claim 1, characterized in that: The identification feature of the hoisting section in step 2 is: the crane is performing hoisting operations and there are objects hoisted; the identification feature of the drum forming section is: the plate rolling machine is in operation and the steel plate is being rolled; the identification feature of the external welding section is: the staff is performing welding operations on the outside of the drum; the identification feature of the internal welding section is: the staff is performing welding operations inside the drum.

3. The method for safety monitoring of the production process of cylinder equipment according to claim 1 or 2 is characterized in that: The specific implementation method of step 2 includes: S21: Extract the videos collected in step 1, including the videos of hoisting operation, plate rolling machine plate rolling, external welding of the roll, and internal welding of the roll; divide the video into a segment every 10 frames, label each segment with a category, and construct an action recognition dataset; S22: extracting target images including a hook, a reel, a human body, goggles, a protective mask, protective gloves, welding overalls, a safety belt, sundries and a fan from the video collected in step 1; annotating the targets in the target images, generating an annotation file including target position and category information, and constructing a target detection dataset; S23: using the action recognition algorithm and the target detection algorithm to train the action recognition dataset and the target detection dataset respectively; S24: Use the trained action recognition model and target detection model to identify which process segment the current operation content belongs to and the security risks existing in the process segment.

4. The method for safety monitoring of the production process of cylinder equipment according to claim 1 is characterized in that: In step 2, if the current process section is the hoisting section, it is detected whether there is a person standing under the hoisted object, specifically including: First, identify the hook and the person, and obtain the coordinate information of the human body detection frame as [ X min-p , Y min-p , X max-p , Y max-p ], the coordinate information of the hook detection frame is [ X min-h , Y min-h , X max-h , Y maxh ]; Secondly, calculate the horizontal Euclidean distance between the center points of the hook detection frame and the human body detection frame. Finally, if the horizontal Euclidean distance d Less than the specified threshold d th , it means that there is a person standing under the hanging object; if the horizontal Euclidean distance d Greater than a specified threshold d th , it means there is no one standing under the suspended object.

5. The method for safety monitoring of the production process of cylinder equipment according to claim 1 is characterized in that: In step 2, if the current process segment is the roll forming segment, check whether the plate rolling machine operator wears goggles; if the current process segment is the external welding segment, check whether there are debris on the roll, whether the welder wears welding work clothes, protective gloves and protective masks, and whether a safety belt is worn for high-altitude work; if it is detected that there are debris on the roll or the welder is not wearing protective equipment, directly output that there is a safety risk; if there is no debris on the roll and the welder is wearing protective equipment, further determine whether there is high-altitude work, and if there is no high-altitude work, output that there is no safety risk; if there is high-altitude work, further determine whether a safety belt is worn, and if not, output that there is a safety risk.

6. The method for safety monitoring of the production process of cylinder equipment according to claim 5 is characterized in that: Identification method of climbing work: S31: Obtain the coordinate information of the human body detection frame as [ X min-p , Y min-p , X max-p , Y max-p ], the coordinate information of the roll detection frame is [ X min-r , Y min-r , X max-r , Y max-r ]; S32: Calculate the height difference between the human body and the center point of the reel ; S33: Calculate the intersection and union ratio of the human body and the roll: , in, ; S P and S R are the areas of the human body detection frame and the roll detection frame respectively; S34: Based on the intersection and union ratio of the human body detection frame and the roll detection frame IOU The height difference between the center point h Determine whether a person is working at height; the conditions for determining whether a person is working at height on the drum are: h > h th and 0 <IOU<IOU th , h th is the height difference threshold between the human body and the center point of the reel detection frame. IOU th It is the intersection and union ratio threshold of the human body and the reel detection frame.

7. The method for safety monitoring of the production process of cylinder equipment according to claim 1 is characterized in that: In step 2, if the current process segment is the internal welding segment, check whether the welder is wearing welding overalls, protective gloves and protective masks, and whether a fan is placed outside the reel to assist ventilation; the wearing of welding overalls, protective gloves and protective masks, and the placement of the fan are detected in parallel, and as long as none of them is detected, a safety alarm is triggered.

8. The method for safety monitoring of the production process of cylinder equipment according to claim 1 is characterized in that: In step three, the risk assessment conditions include: people standing under the hanging objects; the plate rolling machine operator not wearing goggles; there are debris on the drum; the welder not wearing welding work clothes, protective gloves and protective masks; the welder not wearing a safety belt when working at height; the welder working inside the drum without fan-assisted ventilation.

9. A safety monitoring system for the production process of cylinder equipment, Its characteristic is that it includes a video acquisition module, a security monitoring module and an alarm module; The video acquisition module is used to acquire the video stream of the production area in real time; The safety monitoring module is used to first identify which process segment of the production process the current operation content belongs to, and then use corresponding algorithms according to different process segments to detect safety risks in the production of the process segment; The alarm module generates an alarm based on the type of event triggered by the security risk.

10. The safety monitoring system for the production process of cylinder equipment according to claim 9 is characterized in that: The safety monitoring module includes a classification submodule and a detection submodule; The classification submodule uses a video understanding algorithm to automatically identify the process segment of the current operation, the hoisting segment, the roll forming segment, the external welding segment or the internal welding segment; The detection submodule calls corresponding detection algorithms for each process segment to identify the security risks of different process segments.

Images