Silking machine trailing abnormity identification method, device and equipment and medium

Abstract

The invention provides a silking machine trailing anomaly identification method, device and equipment and a medium, and the method comprises the steps: obtaining a sample image in the production process of a silking machine, marking the sample image to form a training sample data set, building a wire detection model based on the data set, and enabling the model to extract image features and position an abnormal region. The method comprises the following steps: acquiring a to-be-detected image of a production field and inputting the to-be-detected image into a trained model in practical application to obtain an anomaly detection result, screening out an abnormal region in a preset target region, and counting spatial and temporal distribution indexes of the effective abnormal region; if the spatial and temporal distribution indexes simultaneously meet the preset alarm conditions, triggering a trailing abnormity alarm so as to take measures to prevent production accidents; the method improves detection precision and response speed, reduces limitation and potential safety hazards of manual observation, and provides guarantee for safe and efficient operation of a production line.

Description

Technical Field

[0001] The present application relates to the technical field of steel smelting, and in particular to a method, device, equipment and medium for identifying tailing anomalies of a laying head. Background Art

[0002] The laying head, a key piece of equipment in the steelmaking and subsequent wire rod production processes, is widely used in high-speed wire rod rolling lines. Its primary function is to lay the high-temperature rolled wire rod in a uniform spiral onto a cooling conveyor belt. However, in actual operation, improperly set rolling parameters, especially after changing coil specifications or adjusting production parameters, can easily lead to abnormal tailing on the laying head.

[0003] The so-called "tailing anomaly" refers to the situation where the wire fails to fall off along the planned trajectory during the spinning process, and instead becomes partially stuck, stretched, or deviates from the normal path. This anomaly not only degrades the surface quality of the wire and affects the product qualification rate, but in severe cases can even cause the rolled piece to break, resulting in steel pile-up accidents, further damaging production equipment and posing a threat to the safety of on-site workers, posing a significant safety hazard.

[0004] Currently, in actual production, the identification of tailing anomalies on the laying head primarily relies on manual observation. Operators use surveillance video or on-site inspections to determine whether an anomaly exists. Upon detection, they notify relevant personnel to perform wire drawing or adjust rolling parameters. However, this method suffers from delayed response, subjective judgment, and a high rate of missed detections. This often results in untimely processing of anomalies. Each anomaly causes over ten minutes of unplanned downtime, severely impacting production continuity and efficiency. Furthermore, with the increasing automation of production lines, traditional methods relying on manual monitoring are no longer able to meet the demands of modern steel companies for intelligent, efficient, and safe production. Summary of the Invention

[0005] In view of the above-mentioned shortcomings of the prior art, the present invention provides a method, device, equipment and medium for identifying abnormal tailing of a spinning machine to solve the above-mentioned technical problems.

[0006] The present invention provides a method for identifying tailing anomalies of a spinning machine, the method comprising: obtaining sample images of a spinning machine production process, and marking wire contours and tailing anomaly areas in the sample images to form a training sample data set; constructing a wire detection model based on a deep neural network based on the training sample data set, the wire detection model being configured to: extract image features, locate abnormal areas, and fuse multi-area detection results; collect images to be detected at the production site and input them into the trained wire detection model to obtain anomaly detection results, and screen effective abnormal areas in the abnormal detection results that are within a preset target area; and calculating temporal and spatial distribution indicators of the effective abnormal areas, and triggering a tailing anomaly alarm if the temporal and spatial distribution indicators simultaneously meet preset alarm conditions.

[0007] In one embodiment of the present invention, sample images of a spinning machine production process are obtained, including: performing frame skipping extraction on a continuous production video of the spinning machine at a preset time to determine an image frame; intercepting a key area of ​​each frame of the image according to predefined device position coordinates, wherein the key area covers the core working range of the spinning machine, and the boundary of the key area is jointly determined by a height direction coordinate interval and a width direction coordinate interval.

[0008] In one embodiment of the present invention, the effective abnormal area located in the preset target area in the abnormal detection result is screened, including: calculating the spatial geometric intersection of the abnormal area bounding box output by the wire detection model and the preset target area bounding box, wherein the preset target area is defined based on the coordinate interval of the key area; if there is any abnormal area whose intersection area with the original abnormal area exceeds a preset threshold, then the abnormal area is retained and determined as a valid abnormal area.

[0009] In one embodiment of the present invention, a wire detection model based on a deep neural network is constructed based on the training sample dataset, including: obtaining unlabeled industrial scene images and constructing a reference sample dataset based on the industrial scene images; performing unsupervised training on the initial model based on the reference sample dataset to optimize the feature extraction capability of the initial model to obtain a pre-trained model; performing supervised training on the pre-trained model based on the training sample dataset, and adjusting the model parameters of the pre-trained model to obtain a trained wire detection model.

[0010] In one embodiment of the present invention, the fusion of multi-region detection results includes: mapping the anomaly positioning results output from different key areas to a unified coordinate system; redundantly merging the detection results of overlapping areas based on the coordinate mapping relationship, and spatially splicing the detection results of non-overlapping areas to generate a global anomaly distribution map covering the entire spinning machine working surface.

[0011] In one embodiment of the present invention, the spatiotemporal distribution index includes a time dimension index and a space dimension index, wherein the time dimension index includes the ratio of the number of frames in which effective abnormal areas appear in multiple consecutive frames of images; the space dimension index includes the total pixel area of ​​the effective abnormal area in a single frame image.

[0012] In one embodiment of the present invention, the preset alarm conditions include: the cumulative total area of ​​the effective abnormal regions exceeds a set area threshold; the proportion of frames in which the effective abnormal regions appear in consecutive image frames exceeds a set frequency threshold.

[0013] The present application provides a device for identifying tail anomalies of a spinning machine, the device comprising: a sample image acquisition module, for acquiring sample images of a spinning machine production process, and marking wire contours and tail anomaly areas in the sample images to form a training sample data set; a model construction module, for constructing a wire detection model based on a deep neural network based on the training sample data set, the wire detection model being configured to: extract image features, locate abnormal areas, and fuse multi-area detection results; a to-be-detected image acquisition module, for acquiring to-be-detected images of a production site and inputting them into a trained wire detection model to obtain anomaly detection results, and screening effective abnormal areas in the abnormal detection results that are within a preset target area; an abnormal alarm module, for counting spatiotemporal distribution indicators of the effective abnormal areas, and triggering a tail anomaly alarm if the spatiotemporal distribution indicators simultaneously meet preset alarm conditions.

[0014] The present application provides an electronic device, which includes: one or more processors; a storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the electronic device implements the above-mentioned method for identifying the tailing anomaly of a spinning machine.

[0015] The present application provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor of a computer, the computer is caused to execute the above-mentioned method for identifying tailing anomalies of a laying head.

[0016] The beneficial effects of the present invention are as follows: The method for identifying tailing anomalies on a spinning machine, firstly, collects sample images from the spinning machine production process and annotates the wire profile and tailing anomaly areas to construct a high-quality training sample dataset, providing an accurate and representative data foundation for subsequent model training, thereby improving the accuracy of anomaly identification. Secondly, the wire detection model constructed based on a deep neural network can effectively extract key features from the image and accurately locate the anomaly area. Furthermore, the model has the ability to integrate detection results from multiple regions, which helps improve detection robustness in complex backgrounds and under different working conditions, reducing the occurrence of false detections and missed detections. Furthermore, in actual application, the method processes and analyzes real-time images of the production site to screen out valid anomaly areas within the preset target area, further improving the pertinence and reliability of anomaly identification and avoiding false alarms caused by interference from irrelevant areas. Finally, the method uses statistical spatiotemporal distribution indicators of valid anomaly areas and combines them with preset alarm conditions for comprehensive judgment. The alarm is triggered only when multiple indicators are simultaneously met, thereby improving the rationality and stability of the alarm logic, avoiding false triggering due to occasional minor anomalies, and ensuring the authenticity and effectiveness of the alarm information. In summary, the present invention realizes the automatic identification and intelligent judgment of the tailing anomaly of the spinning machine. Compared with the traditional method relying on manual observation, it greatly improves the timeliness and accuracy of anomaly identification, reduces the downtime caused by untimely anomaly handling, improves production efficiency, and reduces the risk of equipment damage.

[0017] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings are incorporated into and constitute a part of the specification, illustrating embodiments consistent with the present application and, together with the specification, serving to explain the principles of the present application. It is obvious that the drawings described below are merely some embodiments of the present application, and a person of ordinary skill in the art can derive other drawings based on these drawings without inventive effort. In the drawings:

[0019] Figure 1 1 is a schematic diagram of an implementation environment of a method for identifying tailing anomalies on a spinning machine, shown in an exemplary embodiment of the present application;

[0020] Figure 2 1 is a flow chart of a method for identifying tailing anomalies of a spinning machine, shown in an exemplary embodiment of the present application;

[0021] Figure 3 1 is a schematic diagram showing abnormal tailing of a spinning machine according to an exemplary embodiment of the present application;

[0022] Figure 4 This is an alarm flow chart of a method for identifying tailing anomalies on a spinning machine, shown in an exemplary embodiment of the present application;

[0023] Figure 5 1 is a block diagram of a device for identifying abnormal tailing of a spinning machine according to an exemplary embodiment of the present application;

[0024] Figure 6 A schematic diagram of the structure of a computer system suitable for implementing an electronic device according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0025] The following describes the embodiments of the present invention with reference to the accompanying drawings and preferred embodiments. Those skilled in the art will readily appreciate the other advantages and benefits of the present invention from the disclosure herein. The present invention may also be implemented or applied through various other specific embodiments, and the various details in this specification may be modified or altered based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are intended only to illustrate the present invention and are not intended to limit the scope of protection of the present invention.

[0026] It should be noted that the illustrations provided in the following embodiments are merely schematic illustrations of the basic concept of the present invention. Therefore, the illustrations only show components related to the present invention and are not drawn according to the number, shape, and size of components in actual implementation. In actual implementation, the type, quantity, and proportion of each component may be changed arbitrarily, and the component layout may also be more complex.

[0027] In the following description, numerous details are discussed to provide a more thorough explanation of the embodiments of the present invention. However, it will be apparent to those skilled in the art that the embodiments of the present invention may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring the embodiments of the present invention.

[0028] Figure 1 It is a schematic diagram of the implementation environment of the method for identifying tailing anomalies of a spinning machine shown in an exemplary embodiment of the present application.

[0029] like Figure 1 As shown, the implementation environment of the method for identifying tailing anomalies of a spinning machine involved in this application includes a data acquisition module 101 and a computer device 102.

[0030] Data acquisition module 101 is used to collect necessary image information from the spinning head production process to support subsequent anomaly detection and analysis. Specifically, this module primarily consists of high-definition cameras, installed in key locations to ensure real-time capture of the wire's motion and shape changes during operation. These cameras not only provide clear images of the wire's outline but also maintain stable image quality under varying lighting conditions, laying the foundation for accurate identification of tailing anomalies. Furthermore, the data acquisition module includes an illumination system to adapt to lighting conditions in diverse production environments and ensure image quality, as well as a synchronization control system to ensure time synchronization between cameras, thereby improving the accuracy of multi-view image fusion.

[0031] The computer device 102 is the core of the entire system, responsible for performing tasks such as data analysis, anomaly determination, and alarm triggering. Its specific functions are as follows: It performs a series of preprocessing operations on the original image from the data acquisition module 101, including but not limited to ROI cropping (extracting the region of interest), noise filtering, and contrast adjustment, so that the subsequent target detection algorithm can work more accurately. The preprocessed image is processed using a pre-trained wire detection model based on a deep neural network. This model has powerful feature extraction capabilities and can accurately locate the tailing anomaly area in the image. It also integrates the detection results of multiple regions of interest to form a comprehensive anomaly assessment. Based on the model inference results, valid anomaly areas within the preset target area are screened, and the spatiotemporal distribution indicators (such as area and frequency) of these areas are calculated. Based on the statistically obtained spatiotemporal distribution indicators, a comprehensive judgment is made in combination with the preset alarm conditions. Only when all conditions are met will the tailing anomaly alarm be triggered. Once the alarm is triggered, detailed alarm information is generated, including key information such as the location coordinates and timestamp of the anomaly area, and is intuitively displayed to the operator through the user interface. It also supports recording alarm events to log files for subsequent analysis and parameter optimization.

[0032] In summary, the present invention's method for identifying tailing anomalies on a spinning machine and its implementation environment enable automated detection and intelligent alarming of tailing anomalies during production, significantly improving production line safety and efficiency. Compared to traditional methods that rely on manual observation, this solution offers higher detection accuracy and faster response times, reducing unplanned downtime and the risk of equipment damage.

[0033] Figure 2 It is a flow chart of a method for identifying tailing anomalies of a spinning machine shown in an exemplary embodiment of the present application.

[0034] like Figure 2 As shown, in an exemplary embodiment, the method for identifying abnormal tailing of a laying machine includes at least steps S210 to S240, which are described in detail as follows:

[0035] Step S210 , obtaining sample images of the production process of the laying head, and marking the wire outline and tailing abnormal area in the sample images to form a training sample data set.

[0036] In one embodiment of the present application, sample images of a spinning machine production process are obtained, including: performing frame skipping extraction on a continuous production video of the spinning machine according to a preset time to determine an image frame; intercepting a key area of ​​each frame of the image according to predefined device position coordinates, wherein the key area covers the core working range of the spinning machine, and the boundary of the key area is jointly determined by a height direction coordinate interval and a width direction coordinate interval.

[0037] In a specific embodiment of the present application, first, image frames are extracted from the monitoring video collected during the continuous operation of the spinning machine at preset time intervals. For example, it is set to extract one frame of image per second, or the sampling period is dynamically adjusted according to the actual operating speed of the production line and the frequency of image changes to ensure that the acquired images are representative and non-redundant. Then, the region of interest (ROI) is cropped for each frame of the image extracted. The cropping operation is performed based on predefined equipment position coordinate information, and the coordinate information is determined by analyzing the spatial layout of the core working area of ​​the spinning machine to ensure that the intercepted key areas can completely cover the main areas of wire spouting, movement, and possible tailing anomalies.

[0038] Specifically, in the image coordinate system, the boundaries of the key region are defined by the starting and ending rows in the height direction, and the starting and ending columns in the width direction. For example, if the total size of the image is H×W×C (where H is the image height, W is the image width, and C is the number of channels), then the key region can be represented by cropping the region of interest from the original image through the ROI. The ROI cropping process can be expressed as:

[0039] ROI=I[y1:y2,x1:x2] Formula (1)

[0040] Where I is the original image, ROI is the cropped region of interest, and I[y1:y2,x1:x2] represents the subregion with row range [y1,y2] and column range [x1,x2] extracted from I. The final cropped ROI region is an image patch of size (y2-y1)×(x2-x1)×C, where (y2-y1) is the height of the ROI region, (x2-x1) is the width of the ROI region, and C is the number of channels.

[0041] It should be noted that the cropped image block retains visual information closely related to the operating status of the spinning machine and removes irrelevant or interfering content in the background, thereby providing a data basis with clear structure and prominent targets for subsequent labeling and model training.

[0042] In summary, the above specific embodiments effectively acquire and streamline the image resources used to construct the training sample data set through interframe sampling combined with ROI cropping technology, improve data processing efficiency, and provide high-quality input for the learning process of the deep neural network model.

[0043] In step S220 , a wire detection model based on a deep neural network is constructed based on the training sample data set. The wire detection model is configured to extract image features, locate abnormal areas, and fuse multi-area detection results.

[0044] In one embodiment of the present application, a wire detection model based on a deep neural network is constructed based on the training sample dataset, including: obtaining unlabeled industrial scene images and constructing a reference sample dataset based on the industrial scene images; performing unsupervised training on the initial model based on the reference sample dataset to optimize the feature extraction capability of the initial model to obtain a pre-trained model; performing supervised training on the pre-trained model based on the training sample dataset, and adjusting the model parameters of the pre-trained model to obtain a trained wire detection model.

[0045] In one specific embodiment of this application, building a wire rod detection model based on a deep neural network first requires collecting a large number of unlabeled videos of spinning machines in various industrial production environments. These videos cover a wide range of possible production conditions and environmental variables, such as varying lighting intensity and background complexity, ensuring that the constructed dataset is broadly representative. Subsequently, frames are extracted from these videos at regular intervals to form an initial reference sample dataset, providing material for subsequent unsupervised training.

[0046] Next, the unlabeled image dataset collected above is used to perform unsupervised pre-training on the initial feature extraction model. At this stage, the main goal is to optimize the model's feature extraction capabilities so that it can autonomously learn how to distinguish different types of visual patterns without manual labeling. By adopting a contrastive learning strategy, the model parameters are adjusted by maximizing the similarity between positive sample pairs and minimizing the difference between negative sample pairs. This method enables the model to identify general features in a variety of production environments, enhancing its adaptability. It should be emphasized that during the unsupervised pre-training of the initial feature extraction model, the contrastive loss is used to optimize its feature extraction capabilities, and its expression is as follows:

[0047]

[0048] in, is a contrast loss function used to measure the similarity between the query feature q and the positive sample feature k+, while considering the negative sample feature k -The influence of ; q is the query feature, representing the feature vector that needs to be matched or recognized; k + is a positive sample feature, a feature vector that belongs to the same category as the query feature q, that is, a feature similar to q; k - is a negative sample feature, a feature vector that does not belong to the same category as the query feature q, that is, a feature that is not similar to q; τ is a temperature parameter used to control the smoothness of the loss function.

[0049] After completing the unsupervised pre-training, the model is fine-tuned in a supervised manner using a training sample dataset containing detailed annotation information (i.e., the wire outline and tailing anomaly areas have been annotated). The model consists of three key parts: feature extraction, tailing anomaly detection, and image stitching. The feature extraction part is responsible for capturing the key features in the input image; the tailing anomaly detection part focuses on identifying specific abnormal areas; and the image stitching part is used to integrate the inference results from multiple regions of interest to form a comprehensive detection conclusion. In particular, a method based on a window multi-head self-attention mechanism is introduced in the feature extraction process. This mechanism allows the model to focus on local details while considering global contextual information, thereby improving its sensitivity to subtle differences in complex scenes. Among them, the expression of the window multi-head self-attention calculation amount is:

[0050] Ω=4hwC 2 +2M 2 hwC Formula (3)

[0051] Among them, M represents the size of the window, h is the height of the feature map, w is the width of the feature map, and C represents the number of channels of the feature map.

[0052] Furthermore, throughout the training process, model performance is continuously monitored, and hyperparameters such as the learning rate and batch size are adjusted as needed. After sufficient iterations, when the model's performance on the validation set stabilizes and meets predetermined criteria, a high-performance wire rod inspection model is considered established. This model not only possesses powerful feature extraction capabilities but also accurately locates potential tailing anomalies. Using a rational image stitching strategy, it generates comprehensive inspection reports, providing strong technical support for the safe operation of the production line.

[0053] Ultimately, through this series of steps—from data preparation, unsupervised pre-training, to supervised fine-tuning—a deep neural network model suitable for detecting tail anomalies on spinning machines was constructed. This process ensured that the model could handle complex industrial environments while accurately performing specific tasks, demonstrating high technical feasibility and practicality. This constructed model significantly improved the detection accuracy and response speed of tail anomalies on spinning machines, reducing downtime caused by untimely anomaly handling, improving production efficiency, and ensuring the safety of equipment and personnel.

[0054] In one embodiment of the present application, the fusion of multi-region detection results includes: mapping the anomaly positioning results output from different key areas to a unified coordinate system; redundantly merging the detection results of overlapping areas based on the coordinate mapping relationship, and spatially splicing the detection results of non-overlapping areas to generate a global anomaly distribution map covering the entire spinning machine working surface.

[0055] In a specific embodiment of the present application, during the operation of a laying head, real-time images are cropped using multiple predefined key regions (ROIs). Each key region corresponds to a specific portion of the laying head's working surface, which together cover the entire working surface. For each key region, the deep neural network model outputs corresponding anomaly localization results.

[0056] Next, the anomaly localization results output by different key regions are mapped to a unified coordinate system. Since each key region has its own independent local coordinate system, it is necessary to convert the detection results within it into the global coordinate system based on the positional relationship of each ROI box relative to the original complete image. For example, if a key region is located in the upper left corner of the original image and its size is w×h, then the position of any point (x′, y′) in the region in the global coordinate system can be expressed as (x+x′, y+y′), where (x, y) is the coordinate of the upper left corner of the key region in the global coordinate system.

[0057] Then, based on the above coordinate mapping relationship, the detection results of the overlapping areas are redundantly merged. In actual operation, since adjacent key areas may have a certain overlap, this may cause the same abnormal area to be detected multiple times. To avoid false positives or repeated alarms caused by this situation, the detection results in these overlapping areas need to be analyzed and merged. Specifically, the intersection over union (IoU) of each abnormal area in the overlapping area can be calculated, and the results with IoU values ​​exceeding a certain threshold are merged, retaining the most representative abnormal area as the final result.

[0058] For non-overlapping areas, the detection results are directly spatially stitched. This means that the detection results within each non-overlapping key area are combined according to their relative positions in the global coordinate system to form a complete global anomaly distribution map covering the entire laying head working surface. This step ensures that even if some anomalies appear in a separate key area, their location distribution across the entire working surface is accurately reflected.

[0059] Finally, after generating the global anomaly distribution map, we further screen out valid anomaly regions within the preset target area (i.e., the core working range of the spinning machine). The method used here is to retain only those detection results that overlap with the preset ROI box range to avoid misidentification caused by external factors (such as equipment shadows, background noise, etc.). The single screening process can be expressed as follows:

[0060] r=r roi ∩r det Formula (4)

[0061] Among them, r roi and r det They represent the reasoning results after the ROI area and the detection model expansion, and r represents the result after screening.

[0062] In step S230 , images to be inspected at the production site are collected and input into the trained wire rod inspection model to obtain abnormality detection results, and valid abnormal areas within the preset target area in the abnormality detection results are screened.

[0063] Figure 3 It is a schematic diagram showing abnormal tailing of a spinning machine according to an exemplary embodiment of the present application.

[0064] In one embodiment of the present application, the effective abnormal area located in the preset target area in the abnormal detection results is screened, including: calculating the spatial geometric intersection of the abnormal area bounding box output by the wire detection model and the preset target area bounding box, the preset target area is defined based on the coordinate interval of the key area; if there is any abnormal area whose intersection area with the original abnormal area exceeds a preset threshold, the abnormal area is retained and determined as a valid abnormal area.

[0065] In one embodiment of the present application, first, based on Figure 3 The image of the spinning machine production site is shown, and the key area coordinate range is determined to define the preset target area. These key areas cover the main working range of the wire, such as the wire outlet and cooling area. Each key area can be defined by the coordinates of its upper left corner and lower right corner, that is, (x start ,y start ) and (x end ,y end ), thereby determining a rectangular bounding box. This bounding box is used to limit the effective range of subsequent anomaly detection, ensuring that only the area related to the core operation of the spinning head is focused.

[0066] Next, for all the abnormal area bounding boxes output by the wire detection model, calculate their spatial geometric intersection with the preset target area bounding box. Assume that the bounding box of a certain abnormal area is represented by (x′ start ,y′ start) and (x′ end ,y′ end ) definition, it is necessary to calculate the intersection area between the two rectangular boxes. The specific steps include: determining the overlapping length of the two rectangular boxes in the X-axis direction and the overlapping length in the Y-axis direction, then calculating the intersection area, and calculating the ratio of the intersection area to the total area of ​​the original abnormal region.

[0067] Furthermore, if the intersection area ratio of any abnormal area and the preset target area exceeds a preset threshold (for example, set to 50%), the abnormal area will be retained and determined as a valid abnormal area. Specifically, for each abnormal area, if the intersection area ratio exceeds the preset threshold, the abnormal area is considered to be located in the preset target area and is valid. All valid abnormal areas that meet the conditions are summarized to form a final list of valid abnormal areas. It should be noted that this step can effectively filter out those false alarm areas that are located in non-critical areas or due to external interference factors, ensuring that only truly relevant abnormalities are identified and processed.

[0068] In a specific embodiment, in a certain frame of image, the wire detection model outputs three abnormal areas, which are recorded as A1, A2, and A3. According to the above steps, the intersection area ratio of each abnormal area and the preset target area is calculated:

[0069] The intersection area of ​​A1 accounts for 70%, which exceeds the preset threshold, so A1 is retained;

[0070] The intersection area of ​​A2 accounts for 30%, which does not reach the threshold, so A2 is excluded;

[0071] The intersection area of ​​A3 accounts for 60%, which also exceeds the threshold, so A3 is retained.

[0072] Ultimately, only A1 and A3 were identified as valid anomaly areas, while A2 was filtered out due to not meeting the criteria. By calculating the spatial geometric intersection of the anomaly area and the preset target area, and determining whether the intersection area exceeds a preset threshold, this method can effectively screen out anomaly areas within the critical operating range, reduce false alarms caused by external interference factors, improve the accuracy and reliability of tailing anomaly identification, and ensure the safe and efficient operation of the production line.

[0073] Step S240 , counting the spatiotemporal distribution indicators of the effective abnormal area, and triggering the tailing abnormality alarm if the spatiotemporal distribution indicators simultaneously meet the preset alarm conditions.

[0074] Figure 4 It is an alarm flow chart of a method for identifying tailing anomalies of a spinning machine shown in an exemplary embodiment of the present application.

[0075] like Figure 4As shown, the image to be inspected is first captured from the production site and fed into a trained wire inspection model to determine whether it contains a tailing anomaly. If so, the total area of ​​the anomaly is further identified and compared to a preset area threshold. If this threshold is exceeded, the algorithm determines whether the proportion of abnormal frames exceeds a preset ratio threshold. If so, an alarm is triggered. Conversely, if there is no tailing anomaly, the total area of ​​the anomaly does not exceed the preset area threshold, or the proportion of abnormal frames does not exceed the preset ratio threshold, the process ends.

[0076] In one embodiment of the present application, the spatiotemporal distribution indicator includes a temporal dimension indicator and a spatial dimension indicator. The temporal dimension indicator includes the ratio of frames containing valid abnormal regions in a continuous multi-frame image, and the spatial dimension indicator includes the total pixel area of ​​the valid abnormal regions in a single frame. Accordingly, the preset alarm conditions include: the cumulative total area of ​​the valid abnormal regions exceeds a set area threshold; and the ratio of frames containing valid abnormal regions in a continuous image exceeds a set frequency threshold.

[0077] In a specific embodiment of the present application, the image to be detected is first detected for abnormalities. If an abnormality is present, a spatial dimension indicator is used to determine whether an alarm should be issued. In this embodiment, the spatial dimension indicator primarily focuses on the total pixel area of ​​the valid abnormal regions within a single frame. After each frame is processed by the anomaly detection model and the valid abnormal regions within the preset target area are screened, the total area of ​​these valid abnormal regions is calculated. Specifically, for each valid abnormal region in each frame, the area enclosed by its bounding box is calculated, and the areas of all valid abnormal regions are added together to obtain the total area of ​​the valid abnormal regions for that frame, Aframe. The temporal dimension indicator focuses on the proportion of frames in a continuous multi-frame image containing valid abnormal regions. By analyzing the continuous multi-frame image, the number of frames containing valid abnormal regions is counted, and the proportion of these frames to the total number of frames is calculated. This proportion can reflect the frequency of abnormality. Assuming that 60 frames are acquired within a time window (e.g., 1 minute), and 40 of them contain valid abnormal regions, the proportion of frames in that time window is 66.67%.

[0078] Based on the above-mentioned spatiotemporal distribution indicators, the preset alarm conditions are set as follows: when the total cumulative area of ​​the effective abnormal regions of all frame images within a certain period of time (such as 1 minute) exceeds the set area threshold, that is, when the tailing abnormal area is larger than the alarm threshold When , it is considered that a tailing anomaly occurs, and its expression is:

[0079]

[0080] in, is the alarm threshold, and alarm is the alarm flag.

[0081] Then, if an abnormal stripping situation is confirmed, the severity of the abnormality is further evaluated. The alarm is triggered only when the number of frames with valid abnormal areas in the continuous image frames exceeds the set frequency threshold. That is, the alarm that meets the requirements is filtered out based on the frequency. The expression is as follows:

[0082]

[0083] Where N is the number of frames hit, T is the time interval between each frame, and t is the time range.

[0084] It should be noted that the method proposed in this embodiment can not only accurately identify the effective abnormal area in a single-frame image, but also comprehensively evaluate the occurrence of abnormalities from two dimensions of time and space, ensuring that an alarm is only issued when the abnormality is indeed serious and occurs frequently, thereby improving the accuracy and reliability of the alarm system and effectively ensuring the safe operation of the production line.

[0085] Figure 5 This is a block diagram of a device for identifying abnormal tailing of a spinning machine, as shown in an exemplary embodiment of the present application. The device can be applied to Figure 1 The device may also be applicable to other exemplary implementation environments and specifically configured in other devices. This embodiment does not limit the implementation environment to which the device is applicable.

[0086] like Figure 5 As shown, the exemplary laying head tail anomaly recognition device includes: a sample image acquisition module 510, a model construction module 520, a to-be-detected image acquisition module 530, and an anomaly alarm module 540.

[0087] Among them, the sample image acquisition module 510 is used to obtain sample images of the spinning machine production process, and mark the wire contour and tailing abnormal area in the sample image to form a training sample data set; the model construction module 520 is used to construct a wire detection model based on a deep neural network based on the training sample data set. The wire detection model is configured to: extract image features, locate abnormal areas, and fuse multi-area detection results; the image acquisition module 530 to be detected is used to collect images to be detected at the production site and input them into the trained wire detection model to obtain abnormal detection results, and screen the effective abnormal areas in the abnormal detection results that are located in the preset target area; the abnormal alarm module 540 is used to count the spatiotemporal distribution indicators of the effective abnormal areas. If the spatiotemporal distribution indicators meet the preset alarm conditions at the same time, the tailing abnormal alarm is triggered.

[0088] It should be noted that the laying head tail anomaly identification device provided in the above-mentioned embodiment and the laying head tail anomaly identification method provided in the above-mentioned embodiment are based on the same concept. The specific manner in which each module and unit performs operations has been described in detail in the method embodiment and will not be repeated here. In actual applications, the laying head tail anomaly identification device provided in the above-mentioned embodiment can, as needed, allocate the above-mentioned functions to different functional modules, that is, divide the internal structure of the device into different functional modules to complete all or part of the functions described above, and this is not limited here.

[0089] An embodiment of the present application also provides an electronic device, comprising: one or more processors; a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the electronic device implements the method for identifying the tailing anomaly of a spinning machine provided in the above-mentioned embodiments.

[0090] Figure 6 The following is a schematic diagram showing the structure of a computer system suitable for implementing an electronic device according to an embodiment of the present application. Figure 6 The computer system 600 of the electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0091] like Figure 6 As shown, the computer system 600 includes a central processing unit (CPU) 601, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 602 or the program loaded from the storage part 608 into the random access memory (RAM) 603, such as executing the method described in the above embodiment. Various programs and data required for system operation are also stored in the RAM 603. The CPU 601, ROM 602 and RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0092] The following components are connected to the I / O interface 605: an input section 606 including a keyboard, a mouse, and the like; an output section 607 including devices such as a cathode ray tube (CRT), a liquid crystal display (LCD), and a speaker; a storage section 608 including a hard disk; and a communication section 609 including a network interface card such as a LAN (Local Area Network) card or a modem. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the I / O interface 605 as needed. Removable media 611, such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, is installed in the drive 610 as needed, so that computer programs read therefrom can be installed into the storage section 608 as needed.

[0093] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 609, and / or installed from a removable medium 611. When the computer program is executed by the central processing unit (CPU) 601, the various functions defined in the system of the present application are executed.

[0094] It should be noted that the computer-readable medium shown in the embodiments of the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium can be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media can include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, which carries a computer-readable computer program. This propagated data signal can take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. A computer program embodied on a computer-readable medium may be transmitted using any suitable medium, including but not limited to wireless, wired, or any suitable combination thereof.

[0095] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. Among them, each box in the flowchart or block diagram can represent a module, program segment, or part of the code, and the above-mentioned module, program segment, or part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0096] The units involved in the embodiments described in this application may be implemented by software or hardware, and the units described may also be set in a processor. In some cases, the names of these units do not constitute limitations on the units themselves.

[0097] Another aspect of the present application provides a computer-readable storage medium having a computer program stored thereon. When executed by a computer processor, the computer program causes the computer to perform the aforementioned method for identifying tailing anomalies in a laying head. The computer-readable storage medium may be included in the electronic device described in the above embodiments, or may exist independently and not be incorporated into the electronic device.

[0098] Another aspect of the present application provides a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the method for identifying tailing anomalies on a laying head provided in each of the above embodiments.

[0099] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the present invention. Anyone skilled in the art may modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, any equivalent modifications or alterations made by one of ordinary skill in the art without departing from the spirit and technical principles disclosed herein are intended to be covered by the claims of the present invention.

Claims

1. A method for identifying abnormal tailing of a laying head, characterized in that: The method comprises: Obtaining sample images of a laying head production process, and marking the wire outline and tailing abnormal area in the sample images to form a training sample data set; Based on the training sample dataset, a wire detection model based on a deep neural network is constructed, wherein the wire detection model is configured to: extract image features, locate abnormal areas, and fuse multi-area detection results; Collect images to be inspected at the production site and input them into the trained wire inspection model to obtain abnormality detection results, and screen the effective abnormal areas in the abnormality detection results that are located in the preset target area; The spatiotemporal distribution indicators of the effective abnormal area are counted, and if the spatiotemporal distribution indicators simultaneously meet the preset alarm conditions, a tailing abnormality alarm is triggered.

2. The method for identifying abnormal tailing of a laying head according to claim 1, characterized in that: Acquire sample images of the laying head production process, including: Perform frame skipping extraction on the continuous production video of the spinning machine according to the preset time to determine the image frame; A key area is intercepted for each frame of image according to predefined device position coordinates. The key area covers the core working range of the spinning machine, and the boundary of the key area is determined by the height direction coordinate interval and the width direction coordinate interval.

3. The method for identifying abnormal tailing of a laying head according to claim 2, characterized in that: Screening the abnormality detection results for valid abnormal areas within a preset target area includes: Calculating the spatial geometric intersection of the abnormal area bounding box output by the wire detection model and the preset target area bounding box, where the preset target area is defined based on the coordinate interval of the key area; If the intersection area of ​​any abnormal area and the original abnormal area exceeds the preset threshold, the abnormal area will be retained and determined as a valid abnormal area.

4. The method for identifying abnormal tailing of a laying head according to claim 1, characterized in that: Based on the training sample dataset, a wire detection model based on a deep neural network is constructed, including: Acquire unlabeled industrial scene images and construct a reference sample dataset based on the industrial scene images; Performing unsupervised training on the initial model based on the reference sample dataset to optimize the feature extraction capability of the initial model to obtain a pre-trained model; The pre-training model is supervisedly trained based on the training sample data set, and the model parameters of the pre-training model are adjusted to obtain a trained wire detection model.

5. The method for identifying abnormal tailing of a laying head according to claim 1, characterized in that: Fusion of multi-region detection results, including: Map the anomaly positioning results output from different key areas to a unified coordinate system; The detection results of overlapping areas are redundantly merged based on the coordinate mapping relationship, and the detection results of non-overlapping areas are spatially spliced ​​to generate a global anomaly distribution map covering the entire spinning machine working surface.

6. The method for identifying tailing anomalies of a laying head according to any one of claims 1 to 5, characterized in that: The spatiotemporal distribution index includes a time dimension index and a space dimension index. The time dimension indicator includes the ratio of frames with valid abnormal areas in a continuous multi-frame image; The spatial dimension indicator includes the total pixel area of ​​the effective abnormal area in a single frame image.

7. The method for identifying tailing anomalies of a laying head according to claim 6, characterized in that: The preset alarm conditions include: The cumulative total area of ​​the effective abnormal area exceeds the set area threshold; The percentage of frames with valid abnormal areas in consecutive image frames exceeds the set frequency threshold.

8. A device for identifying abnormal tailing of a spinning machine, characterized in that: The device comprises: A sample image acquisition module is used to obtain sample images of the laying head production process and mark the wire contour and tailing abnormal area in the sample image to form a training sample data set; A model building module is used to build a wire detection model based on a deep neural network based on the training sample data set, wherein the wire detection model is configured to: extract image features, locate abnormal areas, and fuse multi-area detection results; The image acquisition module to be detected is used to collect images to be detected at the production site and input them into the trained wire detection model to obtain abnormality detection results and screen the effective abnormal areas in the abnormality detection results that are located in the preset target area; The abnormality alarm module is used to count the spatiotemporal distribution indicators of the effective abnormal area, and trigger the tailing abnormality alarm if the spatiotemporal distribution indicators meet the preset alarm conditions at the same time.

9. An electronic device, characterized in that: The electronic device comprises: one or more processors; A storage device for storing one or more programs, which, when executed by the one or more processors, enables the electronic device to implement the method for identifying tailing anomalies of a spinning machine as described in any one of claims 1 to 9.

10. A computer-readable storage medium, characterized in that A computer program is stored thereon, and when the computer program is executed by a processor of a computer, the computer is caused to execute the method for identifying tailing abnormality of a laying machine according to any one of claims 1 to 9.

Images