AI visual intrusion protection alarm system and method for intelligent production line

Abstract

The invention relates to the technical field of intrusion prevention and alarm, and particularly discloses an AI visual intrusion prevention and alarm system and method for an intelligent production line, and the method comprises the following steps: S1, analyzing the historical movement track data of equipment in a production region and the transportation path of a transportation robot, screening out risk equipment which may threaten the transportation process; s2, image acquisition points are arranged on the conveying robot, the front environment is shot in real time, risk equipment is recognized, and the moving range of the conveying robot is predicted based on historical data; s3, judging whether the risk equipment can avoid or not according to the moving range of the risk equipment; if so, planning an avoidance path; if not, the speed of the conveying robot is adjusted, voice prompt is triggered, and safe operation is ensured; according to the method, AI visual identification, track prediction and dynamic path planning are combined, so that the risk of equipment collision in the intelligent production line is effectively reduced, and the transportation efficiency and safety are improved.

Description

Technical Field

[0001] The present invention relates to the field of intrusion protection alarm technology, and in particular to an AI visual intrusion protection alarm system and method for an intelligent production line. Background Art

[0002] In today's intangible cultural heritage lacquerware production sector, with the continuous advancement of technology, flexible intelligent production lines are gradually being applied. These lines aim to improve production efficiency and ensure product quality while minimizing human intervention, achieving automated and intelligent production.

[0003] On the flexible and intelligent production line for intangible cultural heritage lacquerware, a robot transporting earthenware xun (a type of ox) plays a crucial role in material transportation. To ensure safe and accurate transportation, the robot is equipped with an advanced computer vision system that monitors the transport path in real time for obstacles. Upon detecting an obstacle, the robot immediately triggers an emergency brake to prevent a collision.

[0004] However, in actual operation, such emergency braking presents new challenges. Even the slightest contact or the inertial force generated by an abrupt stop can easily damage the oxen. As a key component of intangible cultural heritage lacquerware, the oxen is complex in its production process and possesses both high artistic and economic value. Damage during transportation not only wastes raw materials but also disrupts the smooth progress of the entire production process, increasing production costs. Summary of the Invention

[0005] The purpose of the present invention is to provide an AI visual intrusion protection alarm system and method for intelligent production lines to solve the following technical problems.

[0006] The purpose of the present invention can be achieved through the following technical solutions: An AI visual intrusion protection alarm method for an intelligent production line includes the following steps: Step S1: Obtain all equipment in the production area, obtain historical movement trajectory data of each equipment, and obtain the transportation path of the transport robot transporting the earthenware ox; based on the transportation path and historical movement trajectory data, screen out risky equipment from all equipment; Step S2: Setting a collection point on the transport robot, the collection point is used to obtain a real-time image of the area in front of the transport robot; identifying a risk device in the image, and predicting the movement range of the risk device based on historical movement trajectory data; Step S3: Based on the moving range of the risk device, determine whether the risk device is an avoidable device; if the risk device is an avoidable device, plan the avoidance path of the transport robot; if the risk device is not an avoidable device, adjust the speed of the transport robot according to the preset acceleration threshold and issue a voice prompt.

[0007] As a further solution of the present invention: the process of acquiring the historical movement trajectory data includes: A monitoring interval is set, and the real-time position coordinates of the device are obtained in real time within the monitoring interval. The real-time position coordinates are sequentially connected to obtain the movement trajectory of the device; and the movement trajectories of all devices constitute historical movement trajectory data.

[0008] As a further solution of the present invention: the screening process of risky devices includes: If the moving trajectory of the device intersects the transport path, the device is recorded as a risky device; if the moving trajectory of the device does not intersect the transport path, the device is recorded as a safe device.

[0009] As a further solution of the present invention, the process of identifying risky devices in the captured image includes: Obtain sample images of all risky devices and safety devices, and annotate them to obtain sample data; establish an initial recognition model based on image recognition technology, input the sample data into the initial recognition model, train the initial recognition model, and obtain a risky device recognition model; identify the risky devices in the captured images through the risky device recognition model.

[0010] As a further solution of the present invention: the process of predicting the movement range of the risky device includes: Obtain the time when the transport robot starts transporting, record it as the start time, and obtain the current time in real time. According to the start time and the current time, obtain the transport time period; obtain all captured images acquired during the transport time period, record them as real-time images, determine the position coordinates of the risk equipment in the production area based on the real-time images, and sequentially obtain the position coordinates of the risk equipment in each real-time image, sequentially connect the position coordinates to obtain the current movement trajectory of the risk equipment; Obtaining the movement trajectory of the risk device from the historical movement trajectory data, matching a movement trajectory segment corresponding to the current movement trajectory from the movement trajectory, recording the segment as a corresponding movement trajectory segment, and intercepting a movement trajectory segment within a preset time length after the corresponding movement trajectory segment, recording the segment as a predicted movement trajectory segment; The last position coordinate on the current moving trajectory is obtained in the production area and recorded as the end coordinate; the predicted trajectory segment is spliced ​​to the predicted trajectory segment, and the predicted trajectory of the risk equipment in the production area is obtained, which is recorded as the moving range of the risk equipment.

[0011] As a further solution of the present invention: the process of setting the preset duration includes: Obtain historical transport data of all transport robots, including the total transport time consumed by all transport robots; obtain the average transport time T of the transport robots based on the historical transport data, and record the transport time period as t, thereby obtaining the preset time Tt.

[0012] As a further solution of the present invention: the process of determining whether the risky device is an avoidable device includes: Obtain the intersection of the movement range and the transport path, obtain the timestamp of the transport robot's arrival at the intersection, recorded as the first timestamp t1, and obtain the timestamp of the risky device's arrival at the intersection, recorded as the second timestamp t2; if |t1-t2|∈t´, where t´ is a preset error threshold, then the risky device is recorded as a dangerous device; otherwise, the risky device is directly recorded as an avoidable device; If the risky equipment is recorded as a dangerous equipment, the area occupied by all equipment in the production area at the first timestamp is obtained and recorded as the occupied area; the transport start point and transport end point of the transport robot are obtained. If there is a connecting line between the transport start point and the transport end point that does not intersect the occupied area, the risky equipment is an avoidable equipment; otherwise, the risky equipment is not an avoidable equipment; In a preferred embodiment of the present invention, the process of planning the avoidance path of the transport robot includes: Obtain all the lines between the transport start point and the transport end point that do not intersect the occupied area and record them as candidate routes; obtain the path length and number of turns of the candidate routes to obtain the recommended degree of the candidate routes TD = K × (Len × Nt) -1 , where K is a preset correction coefficient and K>0, Len is the path length of the candidate route, and Nt is the number of turns of the candidate route; the candidate route with the highest recommendation degree is selected as the avoidance path of the transport robot.

[0013] As a further solution of the present invention: the process of adjusting the speed of the transport robot includes: According to the first timestamp and the second timestamp, the set {t1, t2} is obtained, and the minimum value in the set is selected and recorded as Min{t1, t2}; the transport speed of the transport robot is obtained, and the shortest deceleration distance of the transport robot is obtained according to the transport speed and the preset acceleration threshold, and the deceleration start time of the transport robot is determined according to the shortest deceleration distance. When the time is at the deceleration start time, the transport robot decelerates at the preset acceleration threshold.

[0014] An AI visual intrusion protection alarm system for intelligent production lines, including: Pre-division module: Obtain all equipment in the production area, obtain historical movement trajectory data of each equipment, and obtain the transportation path of the transport robot transporting the earthen ocarinas; based on the transportation path and historical movement trajectory data, screen out risky equipment from all equipment; Movement prediction module: Set up collection points on the delivery robot, the collection points are used to obtain real-time images of the front of the delivery robot; identify risk devices in the images, and predict the movement range of the risk devices based on historical movement trajectory data; Protection and alarm module: Determine whether the risky device is an avoidable device based on its movement range; if so, plan the avoidance path of the transport robot; if not, adjust the speed of the transport robot according to the preset acceleration threshold and issue a voice prompt.

[0015] Beneficial effects of the present invention: The AI ​​visual intrusion protection alarm system and method for intelligent production lines provided by the present invention are aimed at the safety protection needs of transport robots transporting earthen ocarinas in the flexible intelligent production line of intangible cultural heritage lacquerware. Through a speed adjustment mechanism based on a preset acceleration threshold, the impact of the violent inertial force generated by traditional emergency braking on the earthen ocarinas is avoided. The system achieves deceleration with a smooth and controllable acceleration by accurately calculating the shortest deceleration distance and the start time of deceleration, ensuring that the earthen ocarinas will not be displaced or damaged due to excessive acceleration during transportation. At the same time, the acceleration threshold is dynamically set based on the critical acceleration value of the earthen ocarinas' quality, adapting to the transportation needs of earthen ocarinas of different qualities, ensuring the integrity of the core components of the intangible cultural heritage lacquerware from a technical level, and reducing the waste of raw materials and increased production costs.

[0016] Furthermore, the present invention utilizes risky equipment screening and movement range prediction mechanisms, enabling the system to proactively identify potentially risky equipment on the transport path, rather than simply responding passively upon encountering obstacles. Through historical trajectory data and real-time image analysis, the current trajectory of risky equipment is accurately matched with historical trajectory fragments, predicting its future movement range and enabling predictive protection. Furthermore, the hierarchical judgment logic for avoidable and unavoidable equipment avoids excessive interference with normal production collaboration scenarios, reduces production line downtime caused by misjudgments, and enhances the intelligent level of safety protection.

[0017] Dynamic path planning for avoidable equipment uses a recommendation algorithm to select the optimal avoidance route, minimizing transport time while ensuring safety. Smooth deceleration control for unavoidable equipment prevents production line interruptions caused by sudden stops and ensures that transport tasks proceed as planned. Furthermore, the integration of the virtual digital twin system with real-time monitoring enables managers to monitor production line status in real time and promptly address abnormal situations, further ensuring the continuity and stability of the production process.

[0018] Furthermore, this system uses real-time Chinese voice broadcasting to provide information such as deceleration status and risk warnings, ensuring on-site operators are promptly informed of production line dynamics and can collaborate effectively, forming a dual security system of intelligent protection and manual intervention. The combination of alarm information and real-time display of production information enables managers to remotely monitor the risk management process, improving the efficiency of safety incident response and promoting the transition of production line safety management from passive response to proactive prevention. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The present invention will be further described below with reference to the accompanying drawings.

[0020] Figure 1 This is a schematic diagram of the steps of an AI visual intrusion protection alarm method for an intelligent production line according to the present invention; Figure 2 This is a functional schematic diagram of an AI visual intrusion protection alarm system for an intelligent production line according to the present invention. DETAILED DESCRIPTION

[0021] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 making any creative efforts shall fall within the scope of protection of the present invention.

[0022] See also Figure 1 As shown, the present invention is an AI visual intrusion protection alarm method for an intelligent production line, comprising the following steps: Step S1: Obtain all equipment in the production area, obtain historical movement trajectory data of each equipment, and obtain the transportation path of the transport robot transporting the earthenware ox; based on the transportation path and historical movement trajectory data, screen out risky equipment from all equipment; Specifically, the production line IoT sensing system is used to accurately collect basic information on all equipment in the production area, including milling machines, collaborative robots, ABB robots, dryers, etc. Basic information includes equipment type, spatial coordinate range, and functional parameters. At the same time, with the help of the equipment motion control module and data storage unit, the complete historical movement trajectory data of each equipment is retrieved. This data records the position change sequence, movement timestamp and other information of the equipment during its past operation cycle. Secondly, based on the production process logic and production line layout, a path planning algorithm generates a transport route for the delivery robot to transport the oxen, specifying its starting point (such as the ox storage station), the nodes along the way (the connection points between the various process operation areas), and the destination (such as the inventory area or the entrance to the next process). Subsequently, a spatial trajectory model is constructed based on the transportation path. The historical movement trajectory data of each device is subjected to spatial overlap analysis and motion time series matching with the model. Devices whose historical trajectories may intersect with the transportation path, and whose intersection period overlaps with the transportation task execution period, are screened out and marked as risky devices, providing targets for subsequent risk prevention and control. As a preferred embodiment of the present invention, the production area is the area occupied by the production line; In a preferred embodiment of the present invention, the process of acquiring the historical movement trajectory data includes: Setting a monitoring interval, obtaining the real-time position coordinates of the device in real time within the monitoring interval, sequentially connecting the real-time position coordinates to obtain the movement trajectory of the device; and forming historical movement trajectory data from the movement trajectories of all devices; Specifically, based on the digital twin model of the production line, a monitoring interval covering the entire production process is first delineated. This interval accurately includes the spatial scope involved in the routine operation of the equipment and the flow of materials. High-precision positioning sensors deployed on the production line, such as UWB positioning base stations and industrial camera visual positioning modules, are used to obtain the three-dimensional real-time position coordinates of the equipment in real time within the selected monitoring interval at an acquisition frequency of ≤50ms. The acquired coordinate sequence is sorted by timestamp, and the real-time position coordinates are connected in sequence through the Bezier curve fitting algorithm to generate a smooth movement trajectory that fits the actual movement law of the equipment. The trajectory data simultaneously records movement parameters such as movement speed, acceleration, and steering angle. All movement trajectories of a single device in a continuous production cycle are integrated, and then the trajectory collection of all devices constitutes a historical movement trajectory database to provide basic data support for risk analysis. The production cycle is such as 1 shift and 100 complete process cycles. In a preferred embodiment of the present invention, the risk device screening process includes: If the moving trajectory of the device intersects the transport path, the device is recorded as a risky device; if the moving trajectory of the device does not intersect the transport path, the device is recorded as a safe device; Specifically, with the help of the production line spatial analysis engine, a three-dimensional spatial intersection operation is performed on the equipment movement trajectory and the transportation path. If the three-dimensional curve of the equipment movement trajectory and the transportation path of the transport robot transporting the earthenware ox have a coordinate overlapping area in the spatial coordinate system, that is, an intersection point, and contain continuous overlapping line segments, and after time axis verification, that is, the equipment trajectory movement period and the transportation task execution period are likely to overlap, then it is determined that there is a risk of interference with the transportation path, and the equipment is recorded as a risky equipment. Conversely, if the equipment trajectory and the transportation path have no three-dimensional spatial intersection, or there is no possibility of time axis overlap, then the equipment is recorded as a safe equipment. At the same time, risky equipment is further classified: if the equipment moves at a high speed at the intersection of the trajectory and the intersection angle with the transport path is ≤30°, it is marked as high-risk equipment; if the equipment moves smoothly at the intersection and the intersection angle is greater than 60°, it is marked as low-risk equipment, providing a basis for subsequent differentiated prevention and control; Step S2: Setting a collection point on the transport robot, the collection point is used to obtain a real-time image of the area in front of the transport robot; identifying a risk device in the image, and predicting the movement range of the risk device based on historical movement trajectory data; Specifically, collection points are rationally set up at key locations such as the front and sides of the transport robot. Using high-definition industrial cameras as collection devices, these collection points can capture real-time images of the transport robot's forward direction and surrounding areas at set intervals, such as 50 milliseconds, to fully cover the robot's forward field of view along its transport path. Using a pre-trained deep learning target recognition model, which is trained on a large amount of production line equipment image data, the deep learning target recognition model can accurately identify the appearance characteristics of various types of equipment, perform intelligent analysis of the captured images, and identify marked risky equipment. At the same time, the historical movement trajectory data corresponding to the risky device is retrieved. Combined with the device's current motion state, such as real-time parameters such as speed, acceleration, and direction of movement, trajectory prediction algorithms, such as the Kalman filter algorithm and the LSTM time series prediction model, are used to predict the risk device's movement range in the short term. This generates a prediction result that includes position range and movement trend, providing data support for subsequent risk response. As a preferred embodiment of the present invention, the acquisition point is based on an image capturing device; In a preferred embodiment of the present invention, the process of identifying risky devices in the captured image includes: Obtain sample images of all risky devices and safety devices, and annotate them to obtain sample data; establish an initial recognition model based on image recognition technology, input the sample data into the initial recognition model, train the initial recognition model, and obtain a risky device recognition model; identify risky devices in the captured images using the risky device recognition model; In a preferred embodiment of the present invention, the process of predicting the movement range of the risky device includes: Obtain the time when the transport robot starts transporting, record it as the start time, and obtain the current time in real time. According to the start time and the current time, obtain the transport time period; obtain all captured images acquired during the transport time period, record them as real-time images, determine the position coordinates of the risk equipment in the production area based on the real-time images, and sequentially obtain the position coordinates of the risk equipment in each real-time image, sequentially connect the position coordinates to obtain the current movement trajectory of the risk equipment; Obtaining the movement trajectory of the risk device from the historical movement trajectory data, matching a movement trajectory segment corresponding to the current movement trajectory from the movement trajectory, recording the segment as a corresponding movement trajectory segment, and intercepting a movement trajectory segment within a preset time length after the corresponding movement trajectory segment, recording the segment as a predicted movement trajectory segment; Obtain the last position coordinate on the current moving trajectory in the production area, which is recorded as the end coordinate; splice the predicted trajectory segment to the predicted trajectory segment, thereby obtaining the predicted trajectory of the risk device in the production area, which is recorded as the moving range of the risk device; The process of setting the preset duration includes: Obtain historical delivery data of all delivery robots, wherein the historical delivery data includes the total delivery time consumed by all delivery robots; obtain an average delivery time T of the delivery robots based on the historical delivery data, and denote the delivery time period as t, thereby obtaining a preset delivery time Tt; Specifically, the trigger signal for the delivery robot to start delivery is obtained and accurately recorded as the start time. The current time is obtained in real time through the production line time synchronization module, and a continuous delivery time period is calculated. Relying on the high-definition camera at the collection point, all images taken during this time period are obtained as real-time images. The production line spatial coordinate mapping algorithm is used to convert the image pixel coordinates into the physical location coordinates of the production area. The location coordinates of the risk equipment are extracted frame by frame, and the current movement trajectory is generated by connecting them in chronological order. The full-cycle trajectory data of the risky equipment is retrieved from the historical movement trajectory database, including trajectories for scenarios such as daily operation and abnormal intervention. The dynamic time warping (DTW) algorithm is used to perform feature matching between the current movement trajectory and the historical trajectory. The matching dimensions include speed sequence, steering angle sequence, and position coordinate sequence. Historical trajectory segments with a similarity of 85% or higher are selected as corresponding trajectory segments. Based on the preset duration of the production line process cycle, the trajectory of that duration is intercepted from the end of the corresponding trajectory segment backward as the predicted trajectory segment, which contains the future movement trend of the equipment. Leveraging the spatial positioning capabilities of the production line's digital twin model, the end coordinates of the current movement trajectory are extracted as the reference point for trajectory splicing. The starting point of the predicted trajectory segment is aligned with the end coordinates, and then smoothly spliced ​​together using a spatial interpolation algorithm to generate a complete predicted trajectory covering the current timeframe to a preset future duration. This trajectory clearly displays the predicted movement range of the risky equipment in the three-dimensional space of the production area. Step S3: Determine whether the risky device is an avoidable device based on its movement range; if so, plan an avoidance path for the transport robot; if not, adjust the speed of the transport robot based on a preset acceleration threshold and issue a voice prompt. Specifically, based on the predicted moving range of the risk equipment, combined with the spatial layout of the production line and safety rules, determine whether the risk equipment is an avoidable equipment; build a spatial conflict model to analyze the spatial overlap and overlapping time between the transport path of the delivery robot and the moving range of the risk equipment; if the interference with the risk equipment can be avoided through path adjustment, it is determined to be an avoidable equipment; at this time, call the path planning algorithm to comprehensively consider the transportation efficiency, soil Xun safety and the operating status of surrounding equipment, and plan an optimal avoidance path. The path must meet the constraints of path length increment ≤ 10% of the original path, turning radius ≥ the minimum safe turning radius of the robot; if the conflict cannot be avoided through path adjustment after analysis, If the risky equipment occupies the only transportation channel and no time difference can be utilized, it will be determined as an unavoidable equipment; then, based on the anti-inertial impact characteristics of the earthenware xun and the production line safety standards, the preset acceleration threshold is called. This threshold is determined through a large number of earthenware xun transportation simulation experiments to ensure that the absolute value of the acceleration does not exceed the critical value of earthenware xun damage; the acceleration required for deceleration is calculated, and the speed of the transport robot is dynamically adjusted to make it slow down or stop smoothly; at the same time, the voice prompt system is triggered to broadcast to the on-site personnel in clear Chinese voice, and the deceleration status and risk information are simultaneously displayed on the production line monitoring interface to ensure that humans and machines can work together to deal with risks; the Chinese voice can be "There is an unavoidable risk on the transportation path, the robot has slowed down, please pay attention to safety"; In a preferred embodiment of the present invention, the process of determining whether the risky device is an avoidable device includes: Obtain the intersection of the movement range and the transport path, obtain the timestamp of the transport robot's arrival at the intersection, recorded as the first timestamp t1, and obtain the timestamp of the risky device's arrival at the intersection, recorded as the second timestamp t2; if |t1-t2|∈t´, where t´ is a preset error threshold, then the risky device is recorded as a dangerous device; otherwise, the risky device is directly recorded as an avoidable device; If the risky equipment is recorded as a dangerous equipment, the area occupied by all equipment in the production area at the first timestamp is obtained and recorded as the occupied area; the transport start point and transport end point of the transport robot are obtained. If there is a connecting line between the transport start point and the transport end point that does not intersect the occupied area, the risky equipment is an avoidable equipment; otherwise, the risky equipment is not an avoidable equipment; In a preferred embodiment of the present invention, the process of planning the avoidance path of the transport robot includes: Obtain all the lines between the transport start point and the transport end point that do not intersect the occupied area and record them as candidate routes; obtain the path length and number of turns of the candidate routes to obtain the recommended degree of the candidate routes TD = K × (Len × Nt) -1 , where K is a preset correction coefficient and K>0, Len is the path length of the candidate route, and Nt is the number of turns of the candidate route; the candidate route with the highest recommendation degree is selected as the avoidance path of the transport robot; In a preferred embodiment of the present invention, the process of adjusting the speed of the transport robot includes: Obtaining a set {t1, t2} based on the first timestamp and the second timestamp, selecting a minimum value in the set and recording it as Min{t1, t2}; obtaining a transport speed of the transport robot, obtaining a shortest deceleration distance of the transport robot based on the transport speed and a preset acceleration threshold, determining a deceleration start time of the transport robot based on the shortest deceleration distance, and when the time is at the deceleration start time, the transport robot decelerates at the preset acceleration threshold; The process of setting the acceleration threshold includes: A calibration experiment was designed, in which several earthen xiao samples of different masses were set, and different acceleration values ​​were set. The earthen xiao samples of different masses were transported at different acceleration values ​​to obtain the critical acceleration value of each earthen xiao sample. The critical acceleration value was the maximum acceleration value at which the earthen xiao sample would not be displaced during transportation. The critical acceleration value of each earthen xiao sample was recorded as sample data, and a regression model was established. The regression model was trained using the sample data to obtain an acceleration setting model. When the transport robot transports the earthen xiao, the mass of the earthen xiao is obtained and input into the acceleration setting model to obtain the critical acceleration value of the earthen xiao, which is recorded as the acceleration threshold; It is worth noting that if there is an object in the real-time image that cannot be identified by the risk device recognition model, it will be directly recorded as a risk device and will not be an avoidable device; and based on each real-time image, the real-time speed and forward direction of the object are obtained, and the motion trajectory of the object is predicted based on the real-time speed and forward direction of the object. If the motion trajectories intersect, deceleration will be started directly at the acceleration threshold, and a voice prompt will be issued.

[0023] An AI visual intrusion protection alarm system for intelligent production lines, including: Pre-division module: Obtain all equipment in the production area, obtain historical movement trajectory data of each equipment, and obtain the transportation path of the transport robot transporting the earthen ocarinas; based on the transportation path and historical movement trajectory data, screen out risky equipment from all equipment; Movement prediction module: Set up collection points on the delivery robot, the collection points are used to obtain real-time images of the front of the delivery robot; identify risk devices in the images, and predict the movement range of the risk devices based on historical movement trajectory data; Protection and alarm module: Determine whether the risky device is an avoidable device based on its movement range; if so, plan the avoidance path of the transport robot; if not, adjust the speed of the transport robot according to the preset acceleration threshold and issue a voice prompt.

[0024] Furthermore, the flexible and intelligent production line of intangible cultural heritage lacquerware in the present invention has multiple functions. The milling machine can be used to complete the polishing of the earthen octagon embryo, and the collaborative robot can carry out the gray base, paint base, and polishing process demonstrations in turn. The ABB robot picks up and places the earthen octagon according to the process and places it in the empty space of the dryer or the product inventory area. After completion, it will be notified by the voice broadcast system; in terms of safety protection, a three-dimensional system of points, lines and surfaces is constructed, with emergency stop protection as the point, grating intrusion protection as the line, and visual protection as the focus of the surface, with Chinese voice real-time alarm; at the same time, it is equipped with a virtual digital twin system to achieve synchronization between the virtual simulation interface and the real production line, and can also display production information in real time, helping the production line to operate efficiently, safely and intelligently.

[0025] The above is a detailed description of one embodiment of the present invention. However, the content described is only a preferred embodiment of the present invention and should not be considered to limit the scope 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 present invention.

Claims

1. An AI visual intrusion protection alarm method for intelligent production lines, characterized in that: The following steps are involved: Step S1: Obtain all equipment in the production area, obtain historical movement trajectory data of each equipment, and obtain the transportation path of the transport robot transporting the earthenware ox; based on the transportation path and historical movement trajectory data, screen out risky equipment from all equipment; Step S2: Setting a collection point on the transport robot, the collection point is used to obtain a real-time image of the area in front of the transport robot; identifying a risk device in the image, and predicting the movement range of the risk device based on historical movement trajectory data; Step S3: Based on the moving range of the risk device, determine whether the risk device is an avoidable device; if the risk device is an avoidable device, plan the avoidance path of the transport robot; if the risk device is not an avoidable device, adjust the speed of the transport robot according to the preset acceleration threshold and issue a voice prompt.

2. The AI ​​visual intrusion protection alarm method for intelligent production lines according to claim 1 is characterized in that: In step S1, the process of acquiring the historical movement trajectory data includes: A monitoring interval is set, and the real-time position coordinates of the device are obtained in real time within the monitoring interval. The real-time position coordinates are sequentially connected to obtain the movement trajectory of the device; and the movement trajectories of all devices constitute historical movement trajectory data.

3. The AI ​​visual intrusion protection alarm method for intelligent production lines according to claim 1 is characterized in that: In step S1, the risk device screening process includes: If the moving trajectory of the device intersects the transport path, the device is recorded as a risky device; if the moving trajectory of the device does not intersect the transport path, the device is recorded as a safe device.

4. The AI ​​visual intrusion protection alarm method for intelligent production lines according to claim 1 is characterized in that: In step S2, the process of identifying risky devices in the captured image includes: Obtain sample images of all risky devices and safety devices, and annotate them to obtain sample data; establish an initial recognition model based on image recognition technology, input the sample data into the initial recognition model, train the initial recognition model, and obtain a risky device recognition model; identify the risky devices in the captured images through the risky device recognition model.

5. The AI ​​visual intrusion protection alarm method for intelligent production lines according to claim 1 is characterized in that: In step S2, the process of predicting the movement range of the risky device includes: Obtain the time when the transport robot starts transporting, record it as the start time, and obtain the current time in real time. According to the start time and the current time, obtain the transport time period; obtain all captured images acquired during the transport time period, record them as real-time images, determine the position coordinates of the risk equipment in the production area based on the real-time images, and sequentially obtain the position coordinates of the risk equipment in each real-time image, sequentially connect the position coordinates to obtain the current movement trajectory of the risk equipment; Obtaining the movement trajectory of the risk device from the historical movement trajectory data, matching a movement trajectory segment corresponding to the current movement trajectory from the movement trajectory, recording the segment as a corresponding movement trajectory segment, and intercepting a movement trajectory segment within a preset time length after the corresponding movement trajectory segment, recording the segment as a predicted movement trajectory segment; The last position coordinate on the current moving trajectory is obtained in the production area and recorded as the end coordinate; the predicted trajectory segment is spliced ​​to the predicted trajectory segment, and the predicted trajectory of the risk equipment in the production area is obtained, which is recorded as the moving range of the risk equipment.

6. The AI ​​visual intrusion protection alarm method for intelligent production lines according to claim 1 is characterized in that: In step S2, the process of setting the preset duration includes: Obtain historical transport data of all transport robots, including the total transport time consumed by all transport robots; obtain the average transport time T of the transport robots based on the historical transport data, and record the transport time period as t, thereby obtaining the preset time Tt.

7. The AI ​​visual intrusion protection alarm method for intelligent production lines according to claim 1 is characterized in that: In step S3, the process of determining whether the risky device is an avoidable device includes: Obtain the intersection of the movement range and the transport path, obtain the timestamp of the transport robot's arrival at the intersection, recorded as the first timestamp t1, and obtain the timestamp of the risky device's arrival at the intersection, recorded as the second timestamp t2; if |t1-t2|∈t´, where t´ is a preset error threshold, then the risky device is recorded as a dangerous device; otherwise, the risky device is directly recorded as an avoidable device; If the risky equipment is recorded as a dangerous equipment, the area occupied by all equipment in the production area at the first timestamp is obtained and recorded as the occupied area; the transport start point and transport end point of the transport robot are obtained. If there is a connecting line between the transport start point and the transport end point that does not intersect the occupied area, the risky equipment is an avoidable equipment; otherwise, the risky equipment is not an avoidable equipment; In a preferred embodiment of the present invention, the process of planning the avoidance path of the transport robot includes: Obtain all the lines between the transport start point and the transport end point that do not intersect the occupied area and record them as candidate routes; obtain the path length and number of turns of the candidate routes to obtain the recommended degree of the candidate routes TD = K × (Len × Nt) -1 , where K is a preset correction coefficient and K>0, Len is the path length of the candidate route, and Nt is the number of turns of the candidate route; the candidate route with the highest recommendation degree is selected as the avoidance path of the transport robot.

8. The AI ​​visual intrusion protection alarm method for intelligent production lines according to claim 7 is characterized in that: In step S3, the process of adjusting the speed of the transport robot includes: According to the first timestamp and the second timestamp, the set {t1, t2} is obtained, and the minimum value in the set is selected and recorded as Min{t1, t2}; the transport speed of the transport robot is obtained, and the shortest deceleration distance of the transport robot is obtained according to the transport speed and the preset acceleration threshold, and the deceleration start time of the transport robot is determined according to the shortest deceleration distance. When the time is at the deceleration start time, the transport robot decelerates at the preset acceleration threshold.

9. An AI visual intrusion protection alarm system for intelligent production lines, characterized in that: include: Pre-division module: Obtain all equipment in the production area, obtain historical movement trajectory data of each equipment, and obtain the transportation path of the transport robot transporting the earthen ocarinas; based on the transportation path and historical movement trajectory data, screen out risky equipment from all equipment; Movement prediction module: Set up collection points on the delivery robot, the collection points are used to obtain real-time images of the front of the delivery robot; identify risk devices in the images, and predict the movement range of the risk devices based on historical movement trajectory data; Protection and alarm module: Determine whether the risky device is an avoidable device based on its movement range; if so, plan the avoidance path of the transport robot; if not, adjust the speed of the transport robot according to the preset acceleration threshold and issue a voice prompt.

Images