Glass fiber filament flying and breaking early warning and disposal method based on AI visual analysis

Abstract

The invention relates to the technical field of AI visual detection, and discloses a glass fiber filament flying and breaking early warning and disposal method based on AI visual analysis, which comprises the following steps: collecting a normal operation video of a coil base, processing the video according to frames to form a plurality of pictures, manually marking the pictures containing broken filaments, and training a model by using the marked pictures to obtain a broken filament early warning and disposal result; the coil base is detected in real time through the model, when it is detected that the number of broken wires reaches a threshold value, the audible and visual alarm of the coil base where the broken wires are located is started, alarm information of the coil base where the broken wires are located is obtained, the alarm information is processed through a definition rule, a hexadecimal instruction is generated, and the instruction is sent to a corresponding worker. And meanwhile, the work number information of the worker is judged through an optical character recognition model.

Description

Technical Field

[0001] The present invention relates to the field of AI visual detection technology, and specifically to a glass fiber flying and broken wire early warning and disposal method based on AI visual analysis. Background Art

[0002] The wire drawing operation in the leak plate workshop is all done manually, and the wire drawing operation is done manually after the flying wire is found. The workers work in shifts and walk back and forth for inspection continuously, which is labor-intensive. Moreover, the leak plate with flying wire cannot be found in time, resulting in delayed disposal operations and production waste. Furthermore, the leak plate temperature is as high as 1200℃, and the radiant heat is high. The diameter of the glass fiber is about 10μm, and it is transparent. It is very difficult to observe whether the glass fiber is broken in the red-hot state, and it is also harmful to the eyes and causes discomfort.

[0003] In order to adjust the thermal balance of the leak plate and improve the continuity of operation, it is necessary to adjust the auxiliary heating current value of the leak plate according to the state and flow of the leak plate flying wire to adjust the temperature. Workers are required to manually record and input information such as the leak plate furnace position, flying wire time, and disposal time to provide initial data for the leak plate temperature adjustment. The labor intensity of workers is high, and there are also cases of mis-input and missed input data.

[0004] The present invention optimizes the target detection algorithm based on YOLO, collects feature pictures of flying and broken fiber in glass fiber, establishes a lightweight target detection method, and embeds the method into an edge AI computing device. The alarm signals of multiple furnace positions are output to the signal receiving device in hexadecimal format through the serial port module to remind workers to perform wire drawing operations, and a complete set of detection and alarm devices are formed in combination with sound and light alarms. The device has the characteristics of high detection accuracy, fast speed, small size, stable operation, etc. Summary of the invention

[0005] The present invention provides a glass fiber flying and broken wire early warning and disposal method based on AI visual analysis, which is used to promote the solution of the problems mentioned in the above background technology.

[0006] The present invention provides the following technical solution: a glass fiber flying and broken wire early warning and disposal method based on AI visual analysis, comprising:

[0007] Get the working area of ​​each stove;

[0008] Each work area is equipped with a wire drawing machine, a worker, multiple cameras and an audible and visual alarm;

[0009] Each staff member wears a signal receiving device;

[0010] The cameras include camera 1 and camera 2;

[0011] One edge AI computing device is used for every three furnaces;

[0012] Number all edge AI computing devices starting from 1.

[0013] Number the furnaces under each edge AI computing device in order from furnace No. 1 to furnace No. 3;

[0014] By collecting video information of the normal operation of the wire drawing machine in each furnace and training the detection model;

[0015] The first step of optimizing the detection model is to use compression and excitation operations;

[0016] The loss function and algorithm are used to optimize the detection model in the second step;

[0017] Use the detection model through the edge AI computing device to detect the operation process of the wire drawing machine in real time, and determine whether to issue an alarm based on the detection results;

[0018] If an alarm is triggered, instructions are generated based on the alarm information and defined rules;

[0019] Conduct inspections on some wire drawing machines based on the alarm information;

[0020] By sending instructions, the staff is notified to conduct on-site inspections and determine whether the staff meets the requirements of the current work area.

[0021] Optionally, the video information of the normal operation of the wire drawing machine in each furnace is collected and the detection model is trained, including:

[0022] Collect video information of the normal operation of the wire drawing machine in each furnace as a sample;

[0023] The sample includes the status events of each wire drawing machine in the furnace during the wire drawing process;

[0024] Status events include normal status and abnormal status;

[0025] Process the samples frame by frame to form multiple sample images;

[0026] Manually label sample images containing abnormal conditions;

[0027] The abnormal state is the wire breaking phenomenon that occurs during the operation of the wire drawing machine;

[0028] The manually annotated sample images are used to train the model, and the model is recorded as the detection model.

[0029] Optionally, the first step of optimizing the detection model by using the compression operation and the excitation operation includes:

[0030] The specific formula for compression operation is as follows:

[0031]

[0032] In the above formula, X represents the feature map, c represents the number of channels of the feature map, H and W represent the height and width of the feature map respectively, Z represents the compressed feature vector, i and j represent the index of the height and width of the feature map in the spatial dimension respectively, and Z c represents the value of the cth channel in the compressed feature vector, X c (i, j) represents the value of the cth channel at the spatial position (i, j) in the original feature map;

[0033] The specific formula for incentive operation is as follows:

[0034] S=F ex (z,W)=σ(W2δ(z,W1z))

[0035] In the above formula, F ex represents the excitation operation, z represents the input vector, and W represents the dimension, including W1 and W 2， W1 and W2 represent the dimensions used to compress and expand the vectors in the fully connected layer, δ represents the activation function of the first fully connected layer, and σ represents the activation function after the second fully connected layer.

[0036] Optionally, the second step of optimizing the detection model using the loss function and algorithm includes:

[0037] Use the loss function EIOU_LOSS for recognition. The loss function formula is as follows:

[0038]

[0039] In the above formula, ELOU_LOSS represents the loss function, IOU represents the ratio of the intersection and union between the predicted box and the true box, ρ represents the Euclidean distance, c is the diagonal length of the minimum closure of the predicted box and the true box, and C w and C h are the width and height of the minimum closure of the predicted box and the true box, b represents the center point of the predicted box, and b gt represents the center point of the real box, w represents the width of the predicted box, and w gt represents the width of the real box, h represents the height of the predicted box, and h gt Indicates the height of the real frame;

[0040] The DIOU_NMS algorithm is used to optimize the model. The DIOU_NMS algorithm formula is as follows:

[0041]

[0042] In the above formula, IOU represents the ratio of the intersection and union between the predicted box and the true box, ρ represents the Euclidean distance, c is the diagonal length of the minimum closure of the predicted box and the true box, b represents the center point of the predicted box, and b g t Indicates the center point of the ground truth box.

[0043] Optionally, the edge AI computing device uses a detection model to perform real-time detection on the operation process of the wire drawing machine, and determines whether to issue an alarm based on the detection result, including:

[0044] Use camera No. 1 to obtain the running video of each furnace wire drawing machine under the edge AI computing device in real time, and record it as the target video;

[0045] Process the target video frame by frame to form multiple target images;

[0046] Start all edge AI computing devices and determine whether they can operate normally;

[0047] If the edge AI computing device cannot operate normally, an alarm will be issued to notify professionals for repair;

[0048] If the edge AI computing device can operate normally, the detection model is used to detect the target image;

[0049] If an abnormal state is found, the target image with the abnormal state is recorded into the abnormal image set;

[0050] Set the abnormal state quantity threshold a;

[0051] If the number of elements in the abnormal image set is less than ah, continue to detect;

[0052] If the number of elements in the abnormal picture set is equal to a, the sound and light alarms corresponding to all elements in the abnormal picture set are activated to sound an alarm.

[0053] Optionally, if an alarm is issued, generating instructions according to the alarm information and the defined rules includes:

[0054] The wire drawing machines corresponding to each element in the abnormal image set are obtained respectively, recorded as abnormal machines, and entered into the abnormal machine set;

[0055] Get the alarm information of each element in the abnormal machine set;

[0056] The alarm information includes the device number of the edge AI computing device to which the abnormal machine belongs, the number of the furnace to which the abnormal machine belongs, the activation status of the abnormal machine switch, and the activation status of the sound and light alarm corresponding to the abnormal machine;

[0057] Process the alarm information of each abnormal machine by defining rules, generate instructions, and enter the instruction set;

[0058] The elements in the instruction set are pushed to the signal receiving devices of the corresponding staff through the API service interface.

[0059] Optionally, the step of processing the alarm information of each abnormal machine by defining rules and generating instructions includes:

[0060] The definition rules are as follows:

[0061] Use a 3-bit binary number to represent all wire drawing machines under the edge AI computing device. From left to right, set the weights of the 3-bit binary numbers to 8, 4, and 2, and set each binary number to 0 or 1 to represent the wire drawing machine number.

[0062] Use 5-bit binary numbers to represent the sound and light alarm and the internal switch of the wire drawing machine;

[0063] The internal switches of the wire drawing machine include switch No. 1, switch No. 2, switch No. 3 and switch No. 4;

[0064] The 5-bit binary numbers represent switch No. 1, switch No. 2, switch No. 3, switch No. 4 and the sound and light alarm from left to right, and each binary number is set to 0 or 1 to represent the internal switch status of the alarm and the wire drawing machine, 0 represents off, and 1 represents on;

[0065] From left to right, set the weights of the 5-bit binary numbers to 1, 8, 4, 2, and 1;

[0066] Take the 3-bit binary number as the first part;

[0067] Use the 5-bit binary number as the second part;

[0068] Combine the first part to the second part in order to form an 8-bit binary number;

[0069] The 8-bit binary number is converted into a hexadecimal number according to the weight of each binary digit, and the hexadecimal number is the instruction.

[0070] Optionally, sending instructions to notify staff to conduct on-site inspection and determining whether the staff meets the requirements of the current work area includes:

[0071] Randomly obtain the working area where the element in the abnormal machine set is located, record it as the target area, and obtain the work number of the corresponding staff member according to the target area, record it as the target work number;

[0072] The work number of the staff in the target area is obtained in real time through the second camera, recorded as the pre-selected work number, and recognized and judged according to the optical character recognition model;

[0073] If the pre-selected worker number is inconsistent with the target worker number, a new instruction is pushed to the signal receiving device of the worker corresponding to the target area;

[0074] If the pre-selected work number is consistent with the target work number, the staff in the target area will be notified to conduct on-site inspection, and the abnormal machines corresponding to the current target area in the abnormal machine set will be eliminated, and the elements in the abnormal machine set will be re-selected and judged;

[0075] If the on-site inspection results are consistent with the alarm information, the exception is handled;

[0076] If the on-site inspection result is inconsistent with the alarm information, the current abnormal machine will be removed from the abnormal machine set;

[0077] If the staff ends the exception handling operation, the optical character recognition model of the current working area is closed.

[0078] Optionally, the inspection operation on some wire drawing machines according to the alarm information includes:

[0079] Enter all wire drawing machines into the machine collection;

[0080] Eliminate all abnormal machine set elements in the machine set;

[0081] Through staff to inspect the elements in the machine collection;

[0082] If an abnormal state is found during the inspection, the abnormal state picture is obtained through the No. 1 camera of the current wire drawing machine and recorded in the inspection set;

[0083] When all elements in the machine set are traversed, the inspection operation ends;

[0084] The detection model is trained using the elements in the inspection set.

[0085] The present invention has the following beneficial effects:

[0086] 1. The glass fiber flying and broken wire warning and disposal method based on AI visual analysis uses AI visual analysis technology to conduct real-time monitoring and early warning of glass fiber flying and broken wires, thereby improving production safety and efficiency. By configuring intelligent equipment and algorithm optimization, it can promptly detect and handle broken wire problems, reducing production interruptions and material waste.

[0087] 2. The glass fiber flying and broken wire warning and disposal method based on AI visual analysis can more accurately identify the operating status of the wire drawing machine by collecting and annotating sample images of abnormal conditions and using sample training detection models, thereby improving the accuracy and reliability of the warning system.

[0088] 3. The glass fiber flying and broken wire warning and disposal method based on AI visual analysis introduces a lightweight ShuffleNetV2 network into the backbone network to reduce the number of parameters and calculations of the backbone network, and embeds the SE attention mechanism into the feature fusion network. The SE attention mechanism includes compression and excitation operations. By learning the importance weight of each channel, the feature map is adaptively recalibrated, so that the network can pay more attention to information-rich channels and suppress less important channels. By assigning different weights to different channels, the feature discrimination ability is enhanced, so that the network can pay more attention to key information when processing images and improve the accuracy of feature recognition.

[0089] 4. The glass fiber flying and broken wire warning and disposal method based on AI visual analysis, by selecting the EIOU_LOSS loss function and the DIOU_NMS algorithm, processes the overlapping area between the predicted box and the true box more finely, has stronger adaptability to the target shape, and has better accuracy and applicability. The EIOU_LOSS loss function takes into account the overlapping area, center distance, and width and height loss. The width and height loss minimizes the difference in width and height between the target box and the anchor box, and the model converges faster, which can reduce the model training cycle, speed up the model training process, and provide more accurate positioning information, thereby improving the accuracy of target detection.

[0090] 5. The glass fiber flying and broken wire warning and disposal method based on AI visual analysis performs real-time detection through edge AI computing equipment, and determines whether the equipment can operate normally, eliminates the probability of production stoppage caused by equipment failure, and defines the alarm timing by setting thresholds. It can quickly respond to abnormal conditions of the wire drawing machine, promptly notify staff for processing, reduce production interruption time, and improve production continuity and product quality.

[0091] 6. The glass fiber flying and broken wire warning and disposal method based on AI visual analysis generates instructions according to the alarm information and defined rules, sends the instructions to the corresponding signal receiving devices through the API service interface, and notifies the staff through the signal receiving devices, which can ensure the timely communication and execution of the instructions and improve the emergency response speed and processing efficiency.

[0092] 7. The glass fiber flying and broken wire warning and disposal method based on AI visual analysis has a clear and standardized format for instructions generated by defined rules, which is easy for staff to understand and execute. It also facilitates the automatic processing and recording of exceptions, improving the accuracy of operations and the traceability of data.

[0093] 8. The glass fiber flying and broken wire warning and disposal method based on AI visual analysis obtains the staff number corresponding to the abnormal machine, uses the signal receiving device and optical character recognition model to confirm the identity of the current staff, ensures that the correct staff handles the abnormal situation, and improves the pertinence and efficiency of the handling.

[0094] 9. The AI-based visual analysis-based early warning and treatment method for glass fiber flying and broken wires can further ensure the normal operation of all wire drawing machines by inspecting non-abnormal machines, timely discover and deal with potential problems, and improve the stability and reliability of the entire production system. At the same time, the abnormal conditions found during the inspection process can be used to further train the detection model and continuously optimize the performance of the model warning.

[0095] 10. The glass fiber flying and broken wire warning and disposal method based on AI visual analysis uses automated equipment to detect flying wire quality, reduces manual work, reduces the difficulty and intensity of workers' work, and improves work efficiency and quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0096] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0097] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0098] Example 1, see Figure 1 , the glass fiber flying and broken wire warning and disposal methods based on AI visual analysis include:

[0099] Get the working area of ​​each stove;

[0100] Each work area is equipped with a wire drawing machine, a worker, multiple cameras and an audible and visual alarm;

[0101] Each staff member wears a signal receiving device;

[0102] The cameras include camera 1 and camera 2;

[0103] One edge AI computing device is used for every three furnaces;

[0104] Number all edge AI computing devices starting from 1.

[0105] Number the furnaces under each edge AI computing device in order from furnace No. 1 to furnace No. 3;

[0106] By collecting video information of the normal operation of the wire drawing machine in each furnace and training the detection model;

[0107] The first step of optimizing the detection model is to use compression and excitation operations;

[0108] The loss function and algorithm are used to optimize the detection model in the second step;

[0109] Use the detection model through the edge AI computing device to detect the operation process of the wire drawing machine in real time, and determine whether to issue an alarm based on the detection results;

[0110] If an alarm is triggered, instructions are generated based on the alarm information and defined rules;

[0111] Conduct inspections on some wire drawing machines based on the alarm information;

[0112] Notify staff to conduct on-site inspections by sending instructions, and determine whether the staff meets the requirements of the current work area;

[0113] Through AI visual analysis technology, real-time monitoring and early warning of glass fiber flying and broken wires are carried out, which improves production safety and efficiency. By configuring intelligent equipment and algorithm optimization, broken wire problems can be discovered and handled in time, reducing production interruptions and material waste.

[0114] The method collects video information of the normal operation of the wire drawing machine in each furnace and trains the detection model, including:

[0115] Collect video information of the normal operation of the wire drawing machine in each furnace as a sample;

[0116] The sample includes the status events of each wire drawing machine in the furnace during the wire drawing process;

[0117] Status events include normal status and abnormal status;

[0118] Process the samples frame by frame to form multiple sample images;

[0119] Manually label sample images containing abnormal conditions;

[0120] The abnormal state is the wire breaking phenomenon that occurs during the operation of the wire drawing machine;

[0121] The manually annotated sample images are used to train the model, and the model is recorded as the detection model;

[0122] By collecting and labeling sample images of abnormal conditions and using the samples to train the detection model, the operating status of the wire drawing machine can be identified more accurately, thereby improving the accuracy and reliability of the early warning system.

[0123] The first step of optimizing the detection model by using compression operation and excitation operation includes:

[0124] The specific formula for compression operation is as follows:

[0125]

[0126] In the above formula, X represents the feature map, c represents the number of channels of the feature map, H and W represent the height and width of the feature map respectively, Z represents the compressed feature vector, i and j represent the index of the height and width of the feature map in the spatial dimension respectively, and Z c represents the value of the cth channel in the compressed feature vector, X c (i, j) represents the value of the cth channel at the spatial position (i, j) in the original feature map;

[0127] The specific formula for incentive operation is as follows:

[0128] S=F ex (z,W)=σ(W2δ(z,W1z))

[0129] In the above formula, F ex represents the excitation operation, z represents the input vector, and W represents the dimension, including W1 and W 2， W1 and W2 represent the dimensions used to compress and expand the vectors in the fully connected layer, δ represents the activation function of the first fully connected layer, and σ represents the activation function after the second fully connected layer;

[0130] A lightweight ShuffleNetV2 network is introduced into the backbone network to reduce the number of parameters and computation of the backbone network. The SE attention mechanism is embedded in the feature fusion network. The SE attention mechanism includes compression and excitation operations. By learning the importance weight of each channel, the feature map is adaptively recalibrated, so that the network can pay more attention to information-rich channels and suppress less important channels. By assigning different weights to different channels, the discriminative ability of features is enhanced, so that the network can pay more attention to key information when processing images and improve the accuracy of feature recognition.

[0131] The second step of optimizing the detection model using the loss function and algorithm includes:

[0132] Use the loss function EIOU_LOSS for recognition. The loss function formula is as follows:

[0133]

[0134] In the above formula, ELOU_LOSS represents the loss function, IOU represents the ratio of the intersection and union between the predicted box and the true box, ρ represents the Euclidean distance, c is the diagonal length of the minimum closure of the predicted box and the true box, and C w and C h are the width and height of the minimum closure of the predicted box and the true box, b represents the center point of the predicted box, and b gt represents the center point of the real box, w represents the width of the predicted box, and w gt represents the width of the real box, h represents the height of the predicted box, and h gt Indicates the height of the real frame;

[0135] The DIOU_NMS algorithm is used to optimize the model. The DIOU_NMS algorithm formula is as follows:

[0136]

[0137] In the above formula, IOU represents the ratio of the intersection and union between the predicted box and the true box, ρ represents the Euclidean distance, c is the diagonal length of the minimum closure of the predicted box and the true box, b represents the center point of the predicted box, and b gt Indicates the center point of the real frame;

[0138] By selecting the EIOU_LOSS loss function and the DIOU_NMS algorithm, the overlapping area between the predicted box and the true box is processed more finely, the adaptability to the target shape is stronger, and the accuracy and applicability are better. The EIOU_LOSS loss function takes into account the overlapping area, center distance, and width and height loss. The width and height loss minimizes the width and height difference between the target box and the anchor box, and the model converges faster. This can reduce the model training cycle, speed up the model training process, and provide more accurate positioning information, thereby improving the accuracy of target detection.

[0139] The edge AI computing device uses the detection model to perform real-time detection on the operation process of the wire drawing machine, and determines whether to issue an alarm based on the detection results, including:

[0140] Use camera No. 1 to obtain the running video of each furnace wire drawing machine under the edge AI computing device in real time, and record it as the target video;

[0141] Process the target video frame by frame to form multiple target images;

[0142] Start all edge AI computing devices and determine whether they can operate normally;

[0143] If the edge AI computing device cannot operate normally, an alarm will be issued to notify professionals for repair;

[0144] If the edge AI computing device can operate normally, the detection model is used to detect the target image;

[0145] If an abnormal state is found, the target image with the abnormal state is recorded into the abnormal image set;

[0146] Set the abnormal state quantity threshold a;

[0147] The abnormal state quantity threshold a is a constant and can be adjusted according to the production situation to define when to issue an alarm later;

[0148] If the number of elements in the abnormal image set is less than ah, continue to detect;

[0149] If the number of elements in the abnormal picture set is equal to a, the sound and light alarms corresponding to all elements in the abnormal picture set are activated to sound an alarm;

[0150] Through edge AI computing devices, real-time detection can be performed to determine whether the equipment can operate normally, eliminate the possibility of production stoppage caused by equipment failure, and define the alarm timing by setting thresholds. It can quickly respond to abnormal conditions of the wire drawing machine, notify staff in time for processing, reduce production interruption time, and improve production continuity and product quality.

[0151] If an alarm is issued, instructions are generated according to the alarm information and the defined rules, including:

[0152] The wire drawing machines corresponding to each element in the abnormal image set are obtained respectively, recorded as abnormal machines, and entered into the abnormal machine set;

[0153] Get the alarm information of each element in the abnormal machine set;

[0154] The alarm information includes the device number of the edge AI computing device to which the abnormal machine belongs, the number of the furnace to which the abnormal machine belongs, the activation status of the abnormal machine switch, and the activation status of the sound and light alarm corresponding to the abnormal machine;

[0155] Process the alarm information of each abnormal machine by defining rules, generate instructions, and enter the instruction set;

[0156] Push the elements in the instruction set to the signal receiving devices of the corresponding staff members through the API service interface;

[0157] Generate instructions based on alarm information and defined rules, send the instructions to the corresponding signal receiving devices through the API service interface, and notify the staff through the signal receiving devices. This can ensure the timely communication and execution of instructions, and improve emergency response speed and processing efficiency.

[0158] The process of processing the alarm information of each abnormal machine by defining rules and generating instructions includes:

[0159] The definition rules are as follows:

[0160] Use a 3-bit binary number to represent all wire drawing machines under the edge AI computing device. From left to right, set the weights of the 3-bit binary numbers to 8, 4, and 2, and set each binary number to 0 or 1 to represent the wire drawing machine number.

[0161] Use 5-bit binary numbers to represent the sound and light alarm and the internal switch of the wire drawing machine;

[0162] The internal switches of the wire drawing machine include switch No. 1, switch No. 2, switch No. 3 and switch No. 4;

[0163] The 5-bit binary numbers represent switch No. 1, switch No. 2, switch No. 3, switch No. 4 and the sound and light alarm from left to right, and each binary number is set to 0 or 1 to represent the internal switch status of the alarm and the wire drawing machine, 0 represents off, and 1 represents on;

[0164] From left to right, set the weights of the 5-bit binary numbers to 1, 8, 4, 2, and 1;

[0165] Take the 3-bit binary number as the first part;

[0166] Use the 5-bit binary number as the second part;

[0167] Combine the first part to the second part in order to form an 8-bit binary number;

[0168] Convert the 8-bit binary number into a hexadecimal number according to the weight of each binary digit, and the hexadecimal number is the instruction;

[0169] The instructions generated by defining rules have a clear and standardized format, which is easy for staff to understand and execute. It also facilitates the automated processing and recording of exceptions, improving the accuracy of operations and the traceability of data.

[0170] The sending of instructions to notify the staff to conduct on-site inspection and determine whether the staff meets the requirements of the current work area includes:

[0171] Randomly obtain the working area where the element in the abnormal machine set is located, record it as the target area, and obtain the work number of the corresponding staff member according to the target area, record it as the target work number;

[0172] The work number of the staff in the target area is obtained in real time through the second camera, recorded as the pre-selected work number, and recognized and judged according to the optical character recognition model;

[0173] If the pre-selected worker number is inconsistent with the target worker number, a new instruction is pushed to the signal receiving device of the worker corresponding to the target area;

[0174] If the pre-selected work number is consistent with the target work number, the staff in the target area will be notified to conduct on-site inspection, and the abnormal machines corresponding to the current target area in the abnormal machine set will be eliminated, and the elements in the abnormal machine set will be re-selected and judged;

[0175] If the on-site inspection results are consistent with the alarm information, the exception is handled;

[0176] If the on-site inspection result is inconsistent with the alarm information, the current abnormal machine will be removed from the abnormal machine set;

[0177] If the staff member ends the exception handling operation, the optical character recognition model of the current working area is closed;

[0178] By obtaining the staff number corresponding to the abnormal machine, using the signal receiving device and optical character recognition model to confirm the identity of the current staff, we can ensure that the correct staff handles the abnormal situation and improve the targetedness and efficiency of the handling.

[0179] The inspection operation of some wire drawing machines according to the alarm information includes:

[0180] Enter all wire drawing machines into the machine collection;

[0181] Eliminate all abnormal machine set elements in the machine set;

[0182] Through staff to inspect the elements in the machine collection;

[0183] If an abnormal state is found during the inspection, the abnormal state picture is obtained through the No. 1 camera of the current wire drawing machine and recorded in the inspection set;

[0184] When all elements in the machine set are traversed, the inspection operation ends;

[0185] Train the detection model using the elements in the inspection set;

[0186] By inspecting non-abnormal machines, we can further ensure the normal operation of all wire drawing machines, timely discover and deal with potential problems, and improve the stability and reliability of the entire production system. At the same time, the abnormal conditions found during the inspection process can be used to further train the detection model and continuously optimize the performance of the model warning.

[0187] By using automated equipment to inspect the quality of flying wires, the amount of manual work can be reduced, the difficulty and intensity of workers' work can be lowered, and work efficiency and quality can be improved.

[0188] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0189] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A glass fiber flying and broken wire early warning and disposal method based on AI visual analysis, characterized by: include, Get the working area of ​​each stove; Each work area is equipped with a wire drawing machine, a signal receiving device, multiple cameras and an audible and visual alarm; Each signal receiving device corresponds to one staff member; The cameras include camera 1 and camera 2; One edge AI computing device is used for every three furnaces; Number all edge AI computing devices starting from 1. Number the furnaces under each edge AI computing device in order from furnace No. 1 to furnace No. 3; By collecting video information of the normal operation of the wire drawing machine in each furnace and training the detection model; The first step of optimizing the detection model is to use compression and excitation operations; The loss function and algorithm are used to optimize the detection model in the second step; Use the detection model through the edge AI computing device to detect the operation process of the wire drawing machine in real time, and determine whether to issue an alarm based on the detection results; If an alarm is triggered, instructions are generated based on the alarm information and defined rules; Conduct inspections on some wire drawing machines based on the alarm information; By sending instructions, the staff is notified to conduct on-site inspections and determine whether the staff meets the requirements of the current work area.

2. The glass fiber flying and broken wire early warning and disposal method based on AI visual analysis according to claim 1 is characterized by: The method collects video information of the normal operation of the wire drawing machine in each furnace and trains the detection model, including: Collect video information of the normal operation of the wire drawing machine in each furnace as a sample; The sample includes the status events of each wire drawing machine in the furnace during the wire drawing process; Status events include normal status and abnormal status; Process the samples frame by frame to form multiple sample images; Manually label sample images containing abnormal conditions; The abnormal state is the wire breaking phenomenon that occurs during the operation of the wire drawing machine; The manually annotated sample images are used to train the model, and the model is recorded as the detection model.

3. The glass fiber flying and broken wire early warning and disposal method based on AI visual analysis according to claim 1 is characterized by: The first step of optimizing the detection model by using compression operation and excitation operation includes: The specific formula for compression operation is as follows: In the above formula, X represents the feature map, c represents the number of channels of the feature map, H and W represent the height and width of the feature map respectively, Z represents the compressed feature vector, i and j represent the index of the height and width of the feature map in the spatial dimension respectively, and Z c represents the value of the cth channel in the compressed feature vector, X c (i, j) represents the value of the cth channel at the spatial position (i, j) in the original feature map; The specific formula for incentive operation is as follows: S=F ex (z,W)=σ(W2δ(z,W1z)) In the above formula, F ex represents the excitation operation, z represents the input vector, and W represents the dimension, including W1 and W 2， W1 and W2 represent the dimensions used to compress and expand the vectors in the fully connected layer, δ represents the activation function of the first fully connected layer, and σ represents the activation function after the second fully connected layer.

4. The glass fiber flying and broken wire early warning and disposal method based on AI visual analysis according to claim 1 is characterized by: The second step of optimizing the detection model using the loss function and algorithm includes: Use the loss function EIOU_LOSS for recognition. The loss function formula is as follows: In the above formula, ELOU_LOSS represents the loss function, IOU represents the ratio of the intersection and union between the predicted box and the true box, ρ represents the Euclidean distance, c is the diagonal length of the minimum closure of the predicted box and the true box, and C w and C h are the width and height of the minimum closure of the predicted box and the true box, b represents the center point of the predicted box, and b gt represents the center point of the real box, w represents the width of the predicted box, and w gt represents the width of the real box, h represents the height of the predicted box, and h gt Indicates the height of the real frame; The DIOU_NMS algorithm is used to optimize the model. The DIOU_NMS algorithm formula is as follows: In the above formula, IOU represents the ratio of the intersection and union between the predicted box and the true box, ρ represents the Euclidean distance, c is the diagonal length of the minimum closure of the predicted box and the true box, b represents the center point of the predicted box, and b g t Indicates the center point of the ground truth box.

5. The glass fiber flying and broken wire early warning and disposal method based on AI visual analysis according to claim 1 is characterized by: The edge AI computing device uses the detection model to perform real-time detection on the operation process of the wire drawing machine, and determines whether to issue an alarm based on the detection results, including: Use camera No. 1 to obtain the running video of each furnace wire drawing machine under the edge AI computing device in real time, and record it as the target video; Process the target video frame by frame to form multiple target images; Start all edge AI computing devices and determine whether they can operate normally; If the edge AI computing device cannot operate normally, an alarm will be issued to notify professionals for repair; If the edge AI computing device can operate normally, the detection model is used to detect the target image; If an abnormal state is found, the target image with the abnormal state is recorded into the abnormal image set; Set the abnormal state quantity threshold a; If the number of elements in the abnormal image set is less than a, continue to detect; If the number of elements in the abnormal picture set is equal to a, the sound and light alarms corresponding to all elements in the abnormal picture set are activated to sound an alarm.

6. The glass fiber flying and broken wire early warning and disposal method based on AI visual analysis according to claim 1 is characterized by: If an alarm is issued, instructions are generated according to the alarm information and the defined rules, including: The wire drawing machines corresponding to each element in the abnormal image set are obtained respectively, recorded as abnormal machines, and entered into the abnormal machine set; Get the alarm information of each element in the abnormal machine set; The alarm information includes the device number of the edge AI computing device to which the abnormal machine belongs, the number of the furnace to which the abnormal machine belongs, the activation status of the abnormal machine switch, and the activation status of the sound and light alarm corresponding to the abnormal machine; Process the alarm information of each abnormal machine by defining rules, generate instructions, and enter the instruction set; The elements in the instruction set are pushed to the signal receiving devices of the corresponding staff through the API service interface.

7. The glass fiber flying and broken wire early warning and disposal method based on AI visual analysis according to claim 6 is characterized by: The process of processing the alarm information of each abnormal machine by defining rules and generating instructions includes: The definition rules are as follows: Use a 3-bit binary number to represent all wire drawing machines under the edge AI computing device. From left to right, set the weights of the 3-bit binary numbers to 8, 4, and 2, and set each binary number to 0 or 1 to represent the wire drawing machine number. Use 5-bit binary numbers to represent the sound and light alarm and the internal switch of the wire drawing machine; The internal switches of the wire drawing machine include switch No. 1, switch No. 2, switch No. 3 and switch No. 4; The 5-bit binary numbers represent switch No. 1, switch No. 2, switch No. 3, switch No. 4 and the sound and light alarm from left to right, and each binary number is set to 0 or 1 to represent the internal switch status of the alarm and the wire drawing machine, 0 represents off, and 1 represents on; From left to right, set the weights of the 5-bit binary numbers to 1, 8, 4, 2, and 1; Take the 3-bit binary number as the first part; Use the 5-bit binary number as the second part; Combine the first part to the second part in order to form an 8-bit binary number; The 8-bit binary number is converted into a hexadecimal number according to the weight of each binary digit, and the hexadecimal number is the instruction.

8. The glass fiber flying and broken wire early warning and disposal method based on AI visual analysis according to claim 1 is characterized by: The sending of instructions to notify the staff to conduct on-site inspection and determine whether the staff meets the requirements of the current work area includes: Randomly obtain the working area where the element in the abnormal machine set is located, record it as the target area, and obtain the work number of the corresponding staff member according to the target area, record it as the target work number; The work number of the staff in the target area is obtained in real time through the second camera, recorded as the pre-selected work number, and recognized and judged according to the optical character recognition model; If the pre-selected worker number is inconsistent with the target worker number, a new instruction is pushed to the signal receiving device of the worker corresponding to the target area; If the pre-selected work number is consistent with the target work number, the staff in the target area will be notified to conduct on-site inspection, and the abnormal machines corresponding to the current target area in the abnormal machine set will be eliminated, and the elements in the abnormal machine set will be re-selected and judged; If the on-site inspection results are consistent with the alarm information, the exception is handled; If the on-site inspection result is inconsistent with the alarm information, the current abnormal machine will be removed from the abnormal machine set; If the staff ends the exception handling operation, the optical character recognition model of the current working area is closed.

9. The glass fiber flying and broken wire early warning and disposal method based on AI visual analysis according to claim 1 is characterized by: The inspection operation of some wire drawing machines according to the alarm information includes: Enter all wire drawing machines into the machine collection; Eliminate all abnormal machine set elements in the machine set; Through staff to inspect the elements in the machine collection; If an abnormal state is found during the inspection, the abnormal state picture is obtained through the No. 1 camera of the current wire drawing machine and recorded in the inspection set; When all elements in the machine set are traversed, the inspection operation ends; The detection model is trained using the elements in the inspection set.

Images