Control method for automatic weld crescent detection in furnace based on video recognition

Abstract

The invention discloses an in-furnace automatic weld crescent detection control method based on video recognition, which comprises a control logic integration system, and a video acquisition module, a model processing system, a front-end display module, an optimization feedback module and a production module which are in communication connection with the control logic integration system, the video acquisition module is in communication connection with the model processing system; by efficiently and accurately monitoring the welding seam crescent operation, the strip steel breakage rate caused by inaccurate tracking of the welding seam in the furnace is effectively reduced, so that the product quality and the yield are improved, and the automation level and the production efficiency are improved.

Description

Technical Field

[0001] The invention relates to the technical field of metal processing and automatic control, and in particular to a control method for automatic welding crescent detection in a furnace based on video recognition. Background Art

[0002] In traditional cold rolling production processes, the monitoring of weld crescents mainly relies on weld detectors and encoders on the rolls; the encoders track the strip position and the weld detectors are used for calibration; these traditional methods have obvious limitations: 1. Weld detector: Due to the limitation of principle and environment, weld detector is usually arranged at the entrance of furnace. It is difficult to use due to the high temperature environment inside the furnace. The length of the furnace is about 500 meters, which leads to the fact that the weld inside the furnace cannot be calibrated for a long time. 2. The problem of elongation of the strip steel in the furnace when heated: The strip steel in the continuous annealing furnace will elongate when heated in the furnace, and the elongation cannot be accurately determined. Since the elongation is difficult to determine, the position of the weld crescent in the furnace is difficult to accurately determine; in order to prevent the strip steel from colliding with the air knife edge baffle when the weld seam leaves the furnace and reaches the zinc pot and the air knife, when the specifications change, a relatively large amount of time is always reserved in advance, such as opening the air knife edge baffle in 10 to 15 seconds; the air knife edge baffle is closed a few seconds after the theoretical weld crescent passes the air knife; Due to human factors and equipment limitations, the above method is difficult to achieve accurate and real-time monitoring and control, resulting in insufficient weld protection, a relatively long time for the edge baffle to be in the open state, and a longer time period for the strip steel before and after the weld when the edge baffle is not in use. These strip steels will have surface quality defects of thickened edges, which need to be cut off at the outlet section of the unit, reducing the product yield rate; if the advance amount of the edge baffle opening is simply adjusted, it is easy to have a weld that widens from narrow, and the strip steel hits the edge baffle, causing the strip steel edge to be damaged or even broken, resulting in equipment damage and unit shutdown for maintenance, causing fluctuations in production stability and product quality and reduced output; Therefore, based on the above technical problems, the present application proposes a control method for automatic weld crescent detection in a furnace based on video recognition, which has a low strip breakage rate and ensures output quality. Summary of the invention

[0003] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a control method for automatic weld crescent detection in a furnace based on video recognition, which has a low strip breakage rate and ensures output quality.

[0004] In order to achieve the above-mentioned purpose, the present invention provides a control method for automatic weld crescent detection in a furnace based on video recognition, comprising a control logic integrated system and a video acquisition module, a model processing system, a front-end display module, an optimization feedback module and a production module that are communicatively connected to the control logic integrated system, wherein the video acquisition module is communicatively connected to the model processing system; the control logic integrated system is used to adjust the production module according to the feedback of the operator or the model processing system and the optimization feedback module, the video acquisition module includes a plurality of cameras set up in the furnace for collecting weld crescent images, the video acquisition module is used to transmit the collected weld crescent images to the model processing system and the front-end display module in real time, the model processing system is used to analyze and learn the model transmitted by the video acquisition module and assist the production module in stable production, the front-end display module is used to display the image transmitted by the video acquisition module, the optimization and feedback module is used for the system to control the production module to adjust according to the real-time monitoring data, and the production module is used to perform the production operation of the weld crescent; The specific method is as follows: S1 The video acquisition module monitors the welding crescent operation in the furnace of the production module in real time and collects the image data of the welding crescent; S2: the video acquisition module transmits the acquired image data to the control logic integration system and the model processing system respectively; S3 The model processing system performs model building and optimization analysis through image data; S4 the model processing system outputs the optimized and analyzed model to the control logic integration system; S5 The control logic integration system adjusts the production module in real time according to the model after optimization analysis; S6: the control logic integration system outputs the model data and the image data to the front-end display module for the operator to monitor the production process and the production optimization method; S7 The operator can make real-time adjustments to the production module according to the model conditions transmitted by the control logic integration system and the operation conditions of the production module; S8 The production module feeds back the production data to the control logic integration system; S9 The control logic integration system organizes the production data and transmits it to the optimization feedback module for model optimization; S10: After the optimization feedback module optimizes the production data model, the optimized model is fed back to the control logic integration system, so that the control logic integration system adjusts the production module.

[0005] Furthermore, the model processing system includes an image processing module, a model selection and training module, and a model deployment and monitoring module. The image processing module is used to optimize the images captured by the video acquisition module and transmit the optimized images to the model selection and training module; the model selection and training module is used to select a suitable deep learning framework and convolutional neural network model for image training and model fusion; the model deployment and real-time monitoring module is used to deploy the trained and optimized model in the production line and identify and locate the position of the weld crescent according to the real-time image in the furnace.

[0006] Furthermore, it also includes a control module, wherein the control module is in communication with the production module; the control module is used for the operator to adjust the production module.

[0007] Furthermore, the front-end display module also includes a plurality of display screens. The front-end display module is in communication connection with the video acquisition module. The display screens are used to present image data optimized and processed by the control logic integration system to the operator.

[0008] Furthermore, the video acquisition module is also communicatively connected with the front-end display module.

[0009] The present invention adopts the above-mentioned solution, and its beneficial effects are: By collecting video data from the camera in the furnace at the weld crescent monitoring point, using image processing and combining it with a deep learning target detection algorithm, the weld crescent is detected in real time on the real-time video screen, connected to the production control system, and linked to the on-site production edge baffle control, the accuracy of edge baffle control is improved and the risk of belt breakage is reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 It is a schematic diagram of the process in this embodiment.

[0011] Figure 2 Schematic diagram of functional modules in this embodiment. DETAILED DESCRIPTION

[0012] In order to facilitate the understanding of the present invention, the present invention is described more fully below with reference to the accompanying drawings. The accompanying drawings provide preferred embodiments of the present invention. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. The purpose of providing these embodiments is to make the disclosure of the present invention more thoroughly and comprehensively understood.

[0013] See attached Figure 1As shown, in this embodiment, a control method for automatic weld crescent detection in a furnace based on video recognition includes a control logic integration system and a video acquisition module, a model processing system, a front-end display module, an optimization feedback module and a production module that are communicatively connected to the control logic integration system, wherein the video acquisition module is communicatively connected to the model processing system; the control logic integration system is used to adjust the production module according to the feedback of the operator or the model processing system and the optimization feedback module, the video acquisition module includes a plurality of cameras set up in the furnace for collecting weld crescent images, the video acquisition module is used to transmit the collected weld crescent images to the model processing system and the front-end display module in real time, the model processing system is used to analyze and learn the model transmitted by the video acquisition module and assist the production module in stable production, the front-end display module is used to display the image transmitted by the video acquisition module, the optimization and feedback module is used for the system to control the production module to adjust according to the real-time monitoring data, and the production module is used to perform the production operation of the weld crescent; The specific method is as follows: The S1 video acquisition module monitors the welding crescent operation in the furnace of the production module in real time and collects the image data of the welding crescent; The S2 video acquisition module transmits the collected image data to the control logic integration system and the model processing system respectively; The S3 model processing system uses image data to build models and optimize analysis; The S4 model processing system outputs the optimized and analyzed model to the control logic integration system; The S5 control logic integration system makes real-time adjustments to production modules based on the optimized analysis model; The S6 control logic integration system outputs model data and image data to the front-end display module for operators to monitor the production process and production optimization methods; The S7 operator can make real-time adjustments to the production module based on the model conditions transmitted by the monitoring control logic integration system and the operation conditions of the production module; The S8 production module feeds production data back to the control logic integration system; The S9 control logic integration system organizes the production data and transmits it to the optimization feedback module for model optimization; After the S10 optimization feedback module optimizes the production data model, the optimized model is fed back to the control logic integration system, so that the control logic integration system adjusts the production module.

[0014] Furthermore, the model processing system includes an image processing module, a model selection and training module, and a model deployment and monitoring module. The image processing module is used to optimize the images collected by the video acquisition module and transmit the optimized images to the model selection and training module; the model selection and training module is used to select a suitable deep learning framework and convolutional neural network model for image training and model fusion; the model deployment and real-time monitoring module is used to deploy the trained and optimized model in the production line and identify and locate the position of the weld crescent according to the real-time image in the furnace; Specifically, the image processing module uses the camera inside the furnace to collect high-definition images of the weld crescent, and marks the working position of the weld crescent as training data. Secondly, in the data preparation and pre-processing stage, the training data set is expanded by using data enhancement technology to improve the generalization ability of the model. For example, the weld crescent image is rotated, scaled, flipped, and other operations are performed to generate more training samples, which is more convenient for subsequent model training and improvement of production operations. The model selection and training module selects appropriate deep learning frameworks and convolutional neural network (such as CNN) models to train the collected image samples until the model reaches high accuracy. Secondly, by using model fusion technology, the prediction results of multiple deep learning models are fused to improve the accuracy of detection. Object detection models such as YOLO series and Faster R-CNN are used, and then the above prediction results are weighted averaged to increase the accuracy of model prediction; The model deployment and real-time monitoring module uses real-time performance optimization technology to perform model pruning and quantization, reduce the model's computational complexity and memory usage, and improve the speed of real-time processing. The successfully trained model is deployed on the production line, and image data from the camera in the furnace is processed in real time to identify and locate the weld crescent, thereby putting the trained model into actual production, improving production efficiency, and reflecting the results of model training.

[0015] Furthermore, it also includes a control module, wherein the control module is communicatively connected with the production module; the control module is used for the operator to adjust the production module, so that the operator can manually intervene in the production module in time according to the judgment of the production situation, so as to avoid damage to the edge of the strip or even strip breakage or equipment damage due to untimely mechanical judgment, thereby reducing the maintenance requirements and improving overall safety.

[0016] Furthermore, the front-end display module also includes several display screens, which are communicatively connected to the video acquisition module. The display screens are used to present image data optimized and processed by the control logic integration system to the operator, making it easier for the operator to monitor production conditions in multiple situations (optimization models, real-time production, and improvement methods, etc.).

[0017] Furthermore, the video acquisition module is also connected to the front-end display module as a backup data input stream to avoid errors in system transmission and cause untimely image transmission, thereby ensuring the safety of the overall operation.

[0018] The above embodiments are only preferred embodiments of the present invention and are not intended to limit the present invention in any form. Any technician familiar with the art who, without departing from the scope of the technical solution of the present invention, uses the above disclosed technical content to make more possible changes and modifications to the technical solution of the present invention, or modifications are all equivalent embodiments of the present invention. Therefore, all equivalent and equivalent changes made according to the ideas of the present invention without departing from the content of the technical solution of the present invention should be covered within the protection scope of the present invention.

Claims

1. A control method for automatic weld crescent detection in a furnace based on video recognition, characterized in that: It includes a control logic integration system and a video acquisition module, a model processing system, a front-end display module, an optimization feedback module and a production module that are communicatively connected to the control logic integration system, wherein the video acquisition module is communicatively connected to the model processing system; the control logic integration system is used to adjust the production module according to the feedback of the operator or the model processing system and the optimization feedback module, the video acquisition module includes a plurality of cameras set up in the furnace for collecting weld crescent images, the video acquisition module is used to transmit the collected weld crescent images to the model processing system and the front-end display module in real time, the model processing system is used to analyze and learn the model transmitted by the video acquisition module and assist the production module in stable production, the front-end display module is used to display the image transmitted by the video acquisition module, the optimization and feedback module is used for the system to control the production module to adjust according to real-time monitoring data, and the production module is used to perform production operations of weld crescents; The specific method is as follows: S1 The video acquisition module monitors the welding crescent operation in the furnace of the production module in real time and collects the image data of the welding crescent; S2: the video acquisition module transmits the acquired image data to the control logic integration system and the model processing system respectively; S3 The model processing system performs model building and optimization analysis through image data; S4 the model processing system outputs the optimized and analyzed model to the control logic integration system; S5 The control logic integration system adjusts the production module in real time according to the model after optimization analysis; S6: the control logic integration system outputs the model data and the image data to the front-end display module for the operator to monitor the production process and the production optimization method; S7 The operator can make real-time adjustments to the production module according to the model conditions transmitted by the control logic integration system and the operation conditions of the production module; S8 The production module feeds back the production data to the control logic integration system; S9 The control logic integration system organizes the production data and transmits it to the optimization feedback module for model optimization; S10: After the optimization feedback module optimizes the production data model, the optimized model is fed back to the control logic integration system, so that the control logic integration system adjusts the production module.

2. The control method for automatic weld crescent detection in a furnace based on video recognition according to claim 1 is characterized in that: The model processing system includes an image processing module, a model selection and training module, and a model deployment and monitoring module. The image processing module is used to optimize the images collected by the video acquisition module and transmit the optimized images to the model selection and training module; the model selection and training module is used to select a suitable deep learning framework and convolutional neural network model for image training and model fusion; the model deployment and real-time monitoring module is used to deploy the trained and optimized model in the production line and identify and locate the position of the weld crescent according to the real-time image in the furnace.

3. The control method for automatic weld crescent detection in a furnace based on video recognition according to claim 1 is characterized in that: It also includes a control module, wherein the control module is in communication with the production module; the control module is used for an operator to adjust the production module.

4. The control method for automatic weld crescent detection in a furnace based on video recognition according to claim 1 is characterized in that: The front-end display module also includes a plurality of display screens. The front-end display module is in communication with the video acquisition module. The display screens are used to present image data optimized and processed by the control logic integration system to the operator.

5. The control method for automatic weld crescent detection in a furnace based on video recognition according to claim 1 is characterized in that: The video acquisition module is also connected in communication with the front-end display module.

Images