Automatic rating method and system for center porosity of continuous casting billet and readable storage medium

Abstract

The invention relates to an automatic rating method and system for center porosity of a continuous casting billet and a readable storage medium. The method comprises the steps that a low-power sample image containing the continuous casting billet to be evaluated is obtained; performing foreground and background segmentation on the low-power sample image by using the trained segmentation model I to obtain a continuous casting billet image, and determining a pixel area of the continuous casting billet; using the trained segmentation identification model 2 to determine whether the continuous casting billet image has center porosity, if yes, segmenting the center porosity to obtain a center porosity image, and determining the number N of effective center porosity and a porosity group region pixel area S2; based on the pixel area S1 of the continuous casting billet and the pixel area S2 of the loose group region, the area proportion P of the central loose group region is determined; and comprehensively rating the low center porosity condition of the continuous casting billet according to the number N of the effective center porosity and the area proportion P of the center porosity group region by taking an industrial standard as a reference. The method improves the recognition efficiency and accuracy of the center porosity of the continuous casting billet.

Description

Technical Field

[0001] The invention relates to the field of image processing and metallurgical technology, and in particular to an automatic rating method, system and readable storage medium for central porosity of a continuous casting billet. Background Art

[0002] Continuous casting technology is one of the core means of modern steel production and a vital link in the steel manufacturing process. As society's requirements for the quality of steel products continue to increase, the quality control of continuous casting billets has become increasingly important. Substandard steel will not only lead to a serious drop in prices, but may even lead to scrapping, causing huge economic losses to steel mills. Therefore, real-time tracking and detection of continuous casting billet quality information has become an indispensable part of the modern steel production process.

[0003] The center porosity of continuous casting billet is a common internal defect. It will have an adverse effect on the mechanical properties, processing performance and service life of steel. For ordinary steel grades, slight center porosity has little effect, but when the center porosity is serious, it will cause a significant decrease in the mechanical properties of the billet thickness direction. In addition, severe center porosity may also cause shear cracks and other problems during subsequent processing, such as cutting thick plates. In the steel production process, the center porosity of the billet needs to be graded, and billets of different grades are used in the production of corresponding steels to improve the efficiency and benefits of the steel plant.

[0004] However, the current evaluation of center porosity is mostly done manually, that is, the rating personnel compare the acid-etched billet image with the rating chart on the metallurgical industry standard YB / T 4003-2016 "Continuous Casting Steel Slab Low-magnification Structural Defect Rating Chart" to obtain the corresponding rating report. However, this manual method has low accuracy and strong subjective will, and cannot fully play the role of low-magnification inspection in continuous casting billet production. In addition, the rating results are stored in the form of reports and cannot be analyzed in a timely manner. Summary of the invention

[0005] The present invention provides an automatic rating method, system and readable storage medium for the central looseness of a continuous casting billet, so as to overcome or at least partially overcome the deficiencies of the prior art.

[0006] In a first aspect, the present application provides an automatic rating method for the central looseness of a continuous casting billet, comprising:

[0007] Acquire a low-magnification sample image of a continuous casting slab to be evaluated, wherein the continuous casting slab is subjected to hot acid corrosion treatment;

[0008] Using the trained segmentation model 1, the low-magnification sample image is segmented into foreground and background to obtain a continuous casting billet image, and the pixel area S1 of the continuous casting billet is determined;

[0009] Using the trained segmentation and recognition model 2, determine whether the continuous casting billet image has central porosity. If so, segment the central porosity to obtain a central porosity image, and determine the number N of effective central porosity and the pixel area S2 of the porosity group region;

[0010] Based on the continuous casting billet pixel area S1 and the loose group area pixel area S2, determining the central loose group area area proportion P;

[0011] Based on the industry standard, the low center porosity of the continuous casting billet is comprehensively rated according to the number N of effective center porosity and the area proportion P of the center porosity group.

[0012] In a second aspect, the present application also provides an automatic rating system for the central looseness of a continuous casting billet, the system comprising:

[0013] Acquisition module: used for acquiring a low-magnification sample image of a continuous casting billet to be evaluated, wherein the continuous casting billet is subjected to hot acid corrosion treatment;

[0014] The first recognition module is used to use the trained segmentation model 1 to perform foreground and background segmentation on the low-magnification sample image to obtain a continuous casting billet image and determine the pixel area S1 of the continuous casting billet;

[0015] The second recognition module is used to determine whether the continuous casting billet image has central looseness by using the trained segmentation recognition model 2, and if so, segment the central looseness to obtain a central looseness image, and determine the number N of effective central looseness and the pixel area S2 of the loose group region;

[0016] A data statistics module, for determining a proportion P of a central loose group area based on the pixel area S1 of the continuous casting billet and the pixel area S2 of the loose group area;

[0017] The rating module is used to comprehensively rate the low center porosity of the continuous casting billet based on the industry standard and the number N of effective center porosity and the area proportion P of the center porosity group area.

[0018] In a third aspect, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned method for automatically grading the central porosity of continuous casting billets when executing the computer program.

[0019] In a fourth aspect, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned method for automatically grading the central porosity of the continuous casting billet are implemented.

[0020] This application can at least achieve the following beneficial effects:

[0021] The present application provides an automatic rating method for the center looseness of continuous casting billets, collects low-magnification sample images of continuous casting billets to be evaluated, wherein the continuous casting billets are treated with hot acid corrosion; the adopted model is trained in advance, wherein the segmentation model 1 is mainly used to identify and segment the continuous casting billets in the low-magnification sample images, thereby obtaining the pixel area S1 of the continuous casting billets, and the segmentation and recognition model 2 is mainly used to identify whether there is center looseness in the continuous casting billet image, if there is, then the number of effective center looseness N and the pixel area S2 of the loose group area are identified, and the center loose group area area ratio P is further determined. Based on relevant industry standards, according to the above-obtained data, the automatic and comprehensive rating of the low center looseness of the continuous casting billet can be realized. The present application overcomes the interference of subjective factors during low-magnification rating, reduces the work intensity of workers, improves the rating efficiency and accuracy, and realizes efficient automation of rating; because there is not necessarily center looseness in the casting billet, and semantic segmentation of center looseness is more time-consuming than identifying center looseness, the segmentation method of center looseness in the present application is to identify first and then segment, which improves the recognition efficiency and recognition accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0023] Figure 1 It is a schematic flow chart of an automatic rating method for central looseness of a continuous casting billet in one embodiment of the present application;

[0024] Figure 2 A schematic diagram of an evaluation function iou provided in an embodiment of the present application;

[0025] Figure 3 A schematic diagram showing the structure of a segmentation and recognition model 2 according to an embodiment of the present application is shown;

[0026] Figure 4 A schematic diagram of the basic architecture of Yolo v3 of the segmentation and recognition model 2 provided in an embodiment of the present application;

[0027] Figure 5 A schematic diagram of the basic architecture of pspnet of the segmentation and recognition model 2 provided in an embodiment of the present application;

[0028] Figure 6 This is a diagram showing the effect of the central loose identification result provided in one embodiment of the present application;

[0029] Figure 7A schematic diagram showing a history query interface of an embodiment of the present application is shown;

[0030] Figure 8 A schematic diagram showing a visualization interface for sample processing according to an embodiment of the present application is shown;

[0031] Fig. 9 A schematic structural diagram of an automatic rating system for center looseness of continuous casting billets according to an embodiment of the present application is shown;

[0032] Fig.10 is a schematic diagram of the structure of a computer device in one embodiment of the present invention;

[0033] Fig.11 It is a schematic diagram of the structure of a computer device in another embodiment of the present invention. DETAILED DESCRIPTION

[0034] 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 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.

[0035] In view of the problem that in the prior art, the rating of center porosity is mostly done manually, which is highly subjective, has low accuracy and low efficiency, the present application provides an automatic rating method for the center porosity of continuous casting billets. Figure 1 is a flow chart of an automatic rating method for the center looseness of a continuous casting billet in one embodiment of the present invention, Figure 1 It can be seen that this embodiment at least includes steps S110 to S150:

[0036] Step S110: Acquire a low-magnification sample image of the continuous casting billet to be evaluated, wherein the continuous casting billet has been subjected to hot acid corrosion treatment.

[0037] Before image acquisition, it is best to perform low-magnification hot acid corrosion treatment on the continuous casting billet sample to be evaluated, and then use a camera to collect low-magnification sample images of the continuous casting billet after hot acid corrosion. Among them, the camera preferably uses a high-precision industrial camera, such as a COMS industrial camera, and the sample image is recommended to use a low-magnification sample image. Low-magnification images usually refer to images observed at a lower magnification in microscopy technology. This type of image can provide an overall view of the sample, which is suitable for quick observation and finding areas of interest. In low-magnification images, the sample details are relatively few, but it can help understand the organizational structure, morphology or distribution of the sample.

[0038] In some embodiments of the present application, the method further includes: recommending low-multiple acid corrosion conditions for the continuous casting billet according to the size and steel type of the continuous casting billet input by the user, so as to achieve quantitative corrosion of the continuous casting billet.

[0039] That is, some embodiments of the present application support users to input the size and steel type of the continuous casting billet to be rated, and the system will recommend the low-magnification corrosion conditions corresponding to the continuous casting billet, achieve quantitative and better corrosion effects, and improve the accuracy of system rating.

[0040] Step S120: Use the trained segmentation model 1 to perform foreground and background segmentation on the low-magnification sample image to obtain a continuous casting billet image, and determine the pixel area S1 of the continuous casting billet.

[0041] The segmentation model 1 is mainly used to identify and segment the continuous casting billet in the low-magnification sample image, so as to obtain the pixel area S1 of the continuous casting billet.

[0042] Segmentation model 1 is trained in advance. In some embodiments of the present application, segmentation model 1 is trained according to the following methods: collecting multiple low-magnification sample images of continuous casting billets, and marking the continuous casting billets in the low-magnification sample images; performing data enhancement operations on the marked low-magnification sample images, and the data enhancement operations include any one or more of horizontal flipping, vertical flipping and translation operations to obtain a data set.

[0043] Each image in the data set is resized and input into the segmentation model 1. The TensorFlow framework is used, binary cross entropy is used as the continuous casting billet recognition loss function of the segmentation model 1, and iou is used as the evaluation criterion. The segmentation model 1 is trained until the convergence requirement is met and the training is stopped; the segmentation model 1 is tested, and the segmentation model 1 whose rating indicator iou is greater than the preset threshold is retained.

[0044] The segmentation model 1 mainly includes the following steps S21 to S25:

[0045] S21: Collect multiple low-magnification sample images of continuous casting billets, manually annotate the continuous casting billets in the low-magnification sample images, and create a training set S Train , test set S Test And the validation set S Val , thus obtaining a data set, the more samples the better.

[0046] S22: Perform data enhancement operations on the prepared data set. In deep learning, it is generally required that the number of samples is sufficient. The more samples there are, the better the trained model effect is and the stronger the generalization ability of the model is. However, since high-quality low-magnification images of continuous casting billets are difficult to obtain and there are fewer samples, it is necessary to perform data enhancement on the low-magnification images of continuous casting billets to improve the quality of the data set. In some embodiments of the present application, the data enhancement operation is performed by selecting one or more of the horizontal flipping, vertical flipping and translation operations of the low-magnification sample image of the continuous casting billet.

[0047] S23: Resize the data set and set each image to the same size. This step can be performed in batches using a script, such as setting the image size to (768, 768, 3), and then sending it to segmentation model one to train segmentation model one.

[0048] S24: During the training process, in the TensorFlow framework, binary cross entropy is used as the loss function for ingot recognition, and iou is used as the evaluation criterion to train the model until the specified number of training times is reached or the loss function is lower than the set value.

[0049] The binary cross entropy formula is shown in formula (1):

[0050]

[0051] Among them, y i is the target true value, y i ∈(0,1); is the model prediction value, Loss is a function that measures the difference between the predicted value and the true value.

[0052] S25: Test the trained model and save the model whose rating indicator iou is greater than the set threshold, otherwise repeat the steps to continue training.

[0053] iou is defined as Figure 2 As shown, the formula is formula (2):

[0054]

[0055] Among them, A is the target area, B is the model prediction area, and iou is used to measure the overlap between the model segmented image and the real image in image segmentation.

[0056] This application does not limit the network structure of the segmentation model 1, and any model that can achieve target recognition and segmentation can be used. In some embodiments, this application recommends the use of a U-net model for construction. In some further embodiments of this application, the U-net model is improved to obtain a segmentation model 1. The improved segmentation model 1 is a U-net model that has been downsampled multiple times, and the activation function is selected as a relu function, and a BN layer (batch normalization layer) is added to the U-net model.

[0057] After training, a trained segmentation model 1 is obtained. The trained segmentation model 1 is used to perform foreground and background segmentation on the low-magnification sample image to obtain a continuous casting billet image, and determine the pixel area S1 of the continuous casting billet.

[0058] Specifically, the low-magnification sample of the continuous casting billet in the collected image is segmented from the background, the foreground is the continuous casting billet, and the equivalent rectangular length pixel number of the continuous casting billet sample is counted as L1, and the equivalent rectangular width pixel number is W1, so as to obtain the pixel area S1 of the continuous casting billet, S1=L1*W1.

[0059] Step S130: using the trained segmentation recognition model 2, determine whether the continuous casting billet image has central porosity. If so, segment the central porosity to obtain at least one central porosity image, and determine the number N of effective central porosity and the pixel area S2 of the loose group region.

[0060] The second segmentation recognition model is mainly used to identify and segment whether there is central porosity in the continuous casting billet image. The second segmentation recognition model is also pre-trained, and the training process is basically the same as the first segmentation model, that is, multiple low-magnification sample images of the continuous casting billet are collected, and the central porosity in the low-magnification sample images are annotated. After each image in the annotated data set is reset to a new size, it is input into the second segmentation recognition model. The TensorFlow framework is used, and the custom loss is used as the central porosity recognition loss function of the second segmentation recognition model. The iou is used as the evaluation standard, and the second segmentation recognition model is trained until the convergence requirement is met and the training is stopped; the second segmentation recognition model is tested, and the second segmentation recognition model whose rating index iou is greater than the preset threshold is retained. That is, the difference from the first segmentation model is that the loss function of the second segmentation recognition model is custom. In some embodiments of the present application, the custom Loss formula is formula (3):

[0061] loss = L box +L obj +L cls Formula (3);

[0062] Among them, L box is the detection error, L obj is the confidence error, L clsis the category error.

[0063] The training process will not be described in detail here.

[0064] The network of the segmentation recognition model 2 and the segmentation model 1 can be the same or different, and this application does not limit this. In some embodiments of this application, it is recommended to use the yolo v3 model and the pspnet model to jointly construct the segmentation recognition model 2. Please refer to Figure 3 , Figure 3 The schematic diagram of the structure of the segmentation recognition model 2 of an embodiment of the present application is shown. Figure 3 It can be seen that the second segmentation recognition model includes the sequentially linked yolo v3 model and pspnet model, among which the yolo v3 model is mainly used for the existence recognition of center looseness, and the pspnet model is mainly used for the segmentation of center looseness. Among them, the structure of the yolo v3 model is as follows Figure 4 As shown, the structure of the pspnet model is as follows Figure 5 As shown, the pspnet model is able to aggregate contextual information from different regions, thereby improving the ability to acquire global information.

[0065] When using Figure 3 The segmentation and recognition model 2 shown determines whether the continuous casting billet image has central porosity. If so, the central porosity is segmented to obtain a central porosity image, including: inputting the continuous casting billet image into the yolo v3 model to identify the existence of central porosity. If so, the continuous casting billet image is input into the pspnet model to segment the central porosity to obtain a central porosity image.

[0066] Specifically, the continuous casting billet image obtained by the above segmentation is sent to the yolo v3 model for category judgment and region of interest extraction, so as to determine whether the continuous casting billet has central porosity, that is, whether central porosity occurs. If not, the whole process ends; if the continuous casting billet contains central porosity, the continuous casting billet image is sent to the pspnet model for semantic segmentation to obtain at least one central porosity image.

[0067] The segmented central loose image is further digitized to obtain the pixel area S2 of the loose group region. S2 is still obtained by equivalent length and equivalent width, which will not be repeated here; and the number of effective central looseness in the central loose image is counted as N. The number of effective central looseness can be one or more, which usually depends on the manufacturing process of the continuous casting billet, etc. In this application, the number of effective central looseness refers to the number of central looseness with effective area.

[0068] Step S140: Determine the area proportion P of the central loose group region based on the pixel area S1 of the continuous casting billet and the pixel area S2 of the loose group region.

[0069] In this application, the area ratio P of the central loose group region is defined as formula (4):

[0070] P = (S2 / S1)*100% Formula (4).

[0071] Step S150: Based on the industry standard, the low center porosity of the continuous casting billet is comprehensively rated according to the number N of effective center porosity and the area proportion P of the center porosity group.

[0072] The system can display the segmentation results and output the number of central porosity and the area ratio of central porosity group. Combined with the metallurgical industry standard YB / T 4003-2016 "Low-magnification structural defect rating diagram of continuous casting steel slab", the system can conduct a comprehensive rating of the selected low-magnification sample images of the continuous casting slab.

[0073] After rating, the rating results are displayed on the interface, such as Figure 6 As shown, the results of the central loose division are displayed in the system's prediction output - loose module, and are easy for users to observe. The loose group area is marked with a box. In this module, users can freely adjust the low-magnification image size to facilitate observation of the central loose segmentation effect. If in doubt, the results can be modified in the rating modification module.

[0074] Depend on Figure 1 It can be seen that the present application provides an automatic rating method for the center looseness of continuous casting billets, collects low-magnification sample images of continuous casting billets to be evaluated, wherein the continuous casting billets are treated with corrosion; the adopted model is trained in advance, wherein the segmentation model 1 is mainly used to identify and segment the continuous casting billets in the low-magnification sample images, thereby obtaining the pixel area S1 of the continuous casting billets, and the segmentation and recognition model 2 is mainly used to identify whether there is center looseness in the continuous casting billet image, if there is, the number of center looseness and the pixel area S2 of the loose group area are identified, and the center loose group area area ratio P is further determined. Based on the relevant industry standards, according to the above-obtained data, the automatic and comprehensive rating of the low center looseness of the continuous casting billet can be realized. The present application overcomes the interference of subjective factors during low-magnification rating, reduces the work intensity of workers, improves the rating efficiency and accuracy, and realizes efficient automation of rating; because there is not necessarily center looseness in the casting billet, and semantic segmentation of center looseness is more time-consuming than identifying center looseness, the segmentation method of center looseness in the present application is to identify first and then segment, which improves the recognition efficiency and recognition accuracy.

[0075] The present application also supports some other functions. In some embodiments, the method also includes: obtaining the casting process information of the continuous casting billet, the casting process information including process parameters and dimensions; recording the casting process information and rating results into a database for users to search and compare the data.

[0076] In some embodiments, the method further includes: in response to the user's selection of at least one of the start and end time, casting speed, steel type, continuous casting machine, and center porosity level, filtering and displaying the historically rated steel types to compare process parameters. That is, the application supports the user to query the rating data generated in the historical time by selecting keywords such as start and end time, casting speed, steel type, continuous casting machine, and center porosity level. The user can compare its process parameters to optimize the continuous casting process.

[0077] like Figure 7 As shown, Figure 7 A schematic diagram of a historical query interface of an embodiment of the present application is shown. The casting process information of the continuous casting ingot together with the rating results are stored in the database. The user can use this interface to query the rating data generated in the historical time by selecting the start and end time, casting speed, steel type, continuous casting machine, and center porosity level as keywords.

[0078] In some embodiments, the method also includes: providing a visualization interface, the visualization interface including: a sample parameter control, a rating result control and a situation analysis control; wherein the sample parameter control is used to display a low-magnification sample image of the continuous casting billet and sample parameter details, wherein the sample parameter details include: billet size, pattern number, pattern name, commission number, pulling speed, command number, steel-out mark, furnace number, rater, and at least one of the overheating information; the rating result control is used to display the rating result of the central porosity of the continuous casting billet; the situation analysis control is used to display the manual or intelligent analysis of the cause of the central porosity of the continuous casting billet.

[0079] Figure 8 A schematic diagram showing a visualization interface for sample processing according to an embodiment of the present application is shown. Figure 8 It can be seen that the visualization interface mainly includes but is not limited to: sample parameters, rating results and situation analysis, etc. The sample parameters are to select the low-magnification sample photos of the continuous casting billet to be evaluated and input the corresponding sample parameters, including billet size, pattern number, pattern name, commission number, drawing speed, command number, steel-out mark, furnace number, rater, overheat information, etc. After the system completes the rating of the input photos, the rating results will be displayed in the rating result part of the module, and the causes of the looseness of the continuous casting billet center will be analyzed, and the evaluation results and sample parameters will be stored in the database.

[0080] In some embodiments, the method further comprises: analyzing the characteristics of central porosity of multiple continuous casting billets in the standard database, and displaying pictures of central porosity at each level for quantitative analysis. Analyzing the characteristics of central porosity in the national standard, and displaying pictures of central porosity at each level, and performing quantitative analysis on them is helpful to improve the casting process.

[0081] In some embodiments, the method further includes: in response to a modification or supplement operation by a user, modifying the data of the central looseness of the continuous casting billet generated in the historical time, that is, the user can modify, supplement, etc. the historical rating data.

[0082] Fig. 9 The schematic diagram of the structure of the automatic rating system for the center looseness of the continuous casting billet according to an embodiment of the present application is shown; Fig. 9 It can be seen that the automatic rating system 900 for the center looseness of the continuous casting billet includes:

[0083] Acquisition module 910: used to acquire a low-magnification sample image of a continuous casting billet to be evaluated, wherein the continuous casting billet is subjected to corrosion treatment;

[0084] The first recognition module 920 is used to perform foreground and background segmentation on the low-magnification sample image using the trained segmentation model 1 to obtain a continuous casting billet image and determine the pixel area S1 of the continuous casting billet;

[0085] The second recognition module 930 is used to determine whether the continuous casting billet image has central looseness by using the trained segmentation recognition model 2, and if so, segment the central looseness to obtain a central looseness image, and determine the number N of effective central looseness and the pixel area S2 of the loose group region;

[0086] A data statistics module 940 is used to determine the area proportion P of the central loose group region based on the pixel area S1 of the continuous casting billet and the pixel area S2 of the loose group region;

[0087] The rating module 950 is used to comprehensively rate the low center porosity of the continuous casting billet based on the industry standard and the number N of effective center porosity and the area proportion P of the center porosity group area.

[0088] In some embodiments of the present application, in the above-mentioned device, the segmentation model 1 and the segmentation recognition model 2 are respectively trained according to the following method: collecting multiple low-magnification sample images of continuous casting billets, and marking the continuous casting billets and central porosity in the low-magnification sample images respectively; performing data enhancement operations on the marked low-magnification sample images, and the data enhancement operations include any one or more of horizontal flipping, vertical flipping and translation operations to obtain a data set; resizing each image in the data set and inputting them into the segmentation model 1 and the segmentation recognition model 2 respectively, using the TensorFlow framework, using binary cross entropy as the continuous casting billet recognition loss function of the segmentation model 1 and using custom loss as the central porosity recognition loss function of the segmentation recognition model 2, and using iou as the evaluation standard, and training the segmentation model 1 and the segmentation recognition model 2 respectively until the convergence requirements are met and the training is stopped; testing the segmentation model 1 and the segmentation recognition model 2 respectively, and retaining the segmentation model 1 and the segmentation recognition model 2 whose rating index iou is greater than the preset threshold.

[0089] In some embodiments of the present application, in the above-mentioned device, the segmentation model 1 is constructed based on the U-net model, the segmentation model 1 is a U-net model that has been downsampled multiple times, the activation function is selected as the relu function, and a batch normalization layer is added to the U-net model.

[0090] In some embodiments of the present application, in the above-mentioned device, the second segmentation and recognition model includes a YOLO V3 model and a PSPNET model linked in sequence; the second recognition module 930 is used to input the continuous casting billet image into the YOLO V3 model to identify the existence of central porosity. If it exists, the continuous casting billet image is input into the PSPNET model to segment the central porosity and obtain at least one central porosity image.

[0091] In some embodiments of the present application, the above-mentioned device also includes: a data recording module, used to obtain the casting process information of the continuous casting billet, the casting process information including process parameters and dimensions; the casting process information and rating results are recorded in a database for users to search and compare the data.

[0092] In some embodiments of the present application, the above-mentioned device also includes: a process recommendation module, which is used to recommend low-multiple acid corrosion conditions for the continuous casting billet according to the size and steel type of the continuous casting billet input by the user, so as to achieve quantitative corrosion of the continuous casting billet.

[0093] In some embodiments of the present application, the above-mentioned device also includes: a sample processing module, which is used to provide a visualization interface, and the visualization interface includes: a sample parameter control, a rating result control and a situation analysis control; wherein, the sample parameter control is used to display the low-magnification sample image of the continuous casting billet and the sample parameter details, wherein the sample parameter details include: billet size, pattern number, pattern name, commission number, pulling speed, command number, steel-out mark, furnace number, rater, and at least one of the overheating information; the rating result control is used to display the rating result of the central porosity of the continuous casting billet; the situation analysis control is used to display the manual or intelligent analysis of the cause of the central porosity of the continuous casting billet.

[0094] In some embodiments of the present application, the above-mentioned device also includes: a standard data analysis module, which is used to analyze the characteristics of central porosity of multiple continuous casting billets in the standard database and display central porosity pictures of each level for quantitative analysis.

[0095] In some embodiments of the present application, the above-mentioned device further includes: a history query module for screening and displaying historically rated steel grades in response to a user's selection of at least one of the start and end time, casting speed, steel grade, continuous casting machine, and center looseness level to compare process parameters;

[0096] In some embodiments of the present application, the above-mentioned device further includes: a data revision module, which is used to modify the data of the center looseness of the continuous casting billet generated in the historical time in response to the user's modification or supplementary operation.

[0097] It should be noted that the above-mentioned automatic rating system for the center looseness of the continuous casting billet can realize the above-mentioned automatic rating method for the center looseness of the continuous casting billet one by one, which will not be repeated here.

[0098] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Fig.10 As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile and / or volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external client via a network connection. When the computer program is executed by the processor, the functions or steps of the service side of an automatic rating method for the center looseness of a continuous casting billet are realized.

[0099] In one embodiment, a computer device is provided. The computer device may be a client, and its internal structure diagram may be as follows: Fig.11 As shown. The computer device includes a processor, a memory, a network interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external server through a network connection. When the computer program is executed by the processor, the functions or steps on the client side of a method for automatically grading the center looseness of a continuous casting billet are realized.

[0100] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the following steps when executing the computer program:

[0101] Acquire a low-magnification sample image containing a continuous casting billet to be evaluated, wherein the continuous casting billet is subjected to corrosion treatment;

[0102] Using the trained segmentation model 1, the low-magnification sample image is segmented into foreground and background to obtain a continuous casting billet image, and the pixel area S1 of the continuous casting billet is determined;

[0103] Using the trained segmentation and recognition model 2, determine whether the continuous casting billet image has central porosity. If so, segment the central porosity to obtain a central porosity image, and determine the number N of effective central porosity and the pixel area S2 of the porosity group region;

[0104] Based on the continuous casting billet pixel area S1 and the loose group area pixel area S2, determining the central loose group area area proportion P;

[0105] Based on the industry standard, the low center porosity of the continuous casting billet is comprehensively rated according to the number N of effective center porosity and the area proportion P of the center porosity group.

[0106] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0107] Acquire a low-magnification sample image containing a continuous casting billet to be evaluated, wherein the continuous casting billet is subjected to corrosion treatment;

[0108] Using the trained segmentation model 1, the low-magnification sample image is segmented into foreground and background to obtain a continuous casting billet image, and the pixel area S1 of the continuous casting billet is determined;

[0109] Using the trained segmentation and recognition model 2, determine whether the continuous casting billet image has central porosity. If so, segment the central porosity to obtain a central porosity image, and determine the number N of effective central porosity and the pixel area S2 of the porosity group region;

[0110] Based on the continuous casting billet pixel area S1 and the loose group area pixel area S2, determining the central loose group area area proportion P;

[0111] Based on the industry standard, the low center porosity of the continuous casting billet is comprehensively rated according to the number N of effective center porosity and the area proportion P of the center porosity group.

[0112] It should be noted that the above functions or steps that can be implemented by the computer-readable storage medium or computer device can refer to the relevant descriptions on the server side and the client side in the aforementioned method embodiment. To avoid repetition, they will not be described one by one here.

[0113] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0114] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0115] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. An automatic rating method for the central looseness of a continuous casting billet, characterized in that: include: Acquire a low-magnification sample image of a continuous casting slab to be evaluated, wherein the continuous casting slab is subjected to hot acid corrosion treatment; Using the trained segmentation model 1, the low-magnification sample image is segmented into foreground and background to obtain a continuous casting billet image, and the pixel area S1 of the continuous casting billet is determined; Using the trained segmentation and recognition model 2, determine whether the continuous casting billet image has central porosity. If so, segment the central porosity to obtain a central porosity image, and determine the number N of effective central porosity and the pixel area S2 of the porosity group region; Based on the continuous casting billet pixel area S1 and the loose group area pixel area S2, determining the central loose group area area proportion P; Based on the industry standard, the low center porosity of the continuous casting billet is comprehensively rated according to the number N of effective center porosity and the area proportion P of the center porosity group.

2. The method according to claim 1, characterized in that The segmentation model 1 and the segmentation recognition model 2 are trained respectively according to the following methods: Collecting a plurality of low-magnification sample images of the continuous casting slab, and marking the continuous casting slab and the center looseness in the low-magnification sample images respectively; Performing a data enhancement operation on the annotated low-magnification sample image, wherein the data enhancement operation includes any one or more of a horizontal flip, a vertical flip, and a translation operation to obtain a data set; After resizing each image in the data set, the images are input into the segmentation model 1 and the segmentation recognition model 2 respectively. The TensorFlow framework is used, binary cross entropy is used as the continuous casting slab recognition loss function of the segmentation model 1, and the custom loss is used as the center loose recognition loss function of the segmentation recognition model 2. The iou is used as the evaluation standard, and the segmentation model 1 and the segmentation recognition model 2 are trained respectively until the convergence requirements are met and the training is stopped; The segmentation model 1 and the segmentation recognition model 2 are tested respectively, and the segmentation model 1 and the segmentation recognition model 2 whose rating index iou is greater than a preset threshold are retained.

3. The method according to claim 1 or 2, characterized in that: The segmentation model 1 is constructed based on the U-net model. The segmentation model 1 is a U-net model that has been downsampled multiple times. The activation function is selected as the relu function, and a batch normalization layer is added to the U-net model.

4. The method according to claim 1 or 2, characterized in that: The second segmentation and recognition model includes a yolo v3 model and a pspnet model linked in sequence; The second trained segmentation recognition model is used to determine whether the continuous casting billet image has central porosity. If so, the central porosity is segmented to obtain a central porosity image, including: The continuous casting billet image is input into the YOLO V3 model to identify the existence of central porosity. If it exists, the continuous casting billet image is input into the PSPNET model to segment the central porosity to obtain a central porosity image.

5. The method according to claim 1, characterized in that The method further comprises: Acquiring the continuous casting process information of the continuous casting billet, wherein the continuous casting process information includes process parameters and dimensions; The casting process information and rating results are recorded in a database for users to search and compare the data.

6. The method according to claim 1, characterized in that The method further comprises: According to the size and steel type of the continuous casting billet input by the user, low-multiple acid corrosion conditions are recommended for the continuous casting billet to achieve quantitative corrosion of the continuous casting billet.

7. The method according to claim 1, characterized in that The method further comprises: Providing a visualization interface, the visualization interface including: a sample parameter control, a rating result control, and a situation analysis control; The sample parameter control is used to display the low-magnification sample image of the continuous casting billet and the sample parameter details, wherein the sample parameter details include: at least one of billet size, pattern number, pattern name, commission number, casting speed, order number, steel-out mark, furnace number, rater, and superheat information; The rating result control is used to display the center porosity rating result of the continuous casting billet; The situation analysis control is used to display manual or intelligent analysis of the cause of central porosity in the continuous casting billet.

8. The method according to claim 1, characterized in that The method further comprises: Analyze the characteristics of center porosity of multiple continuous casting billets in the standard database and display center porosity pictures at each level for quantitative analysis; and / or, In response to a user selecting at least one of start and end time, casting speed, steel grade, continuous casting machine, and center porosity level, filtering and displaying historically rated steel grades to compare process parameters; and / or, In response to the modification or supplement operation of the user, the data of the center looseness of the continuous casting billet generated in the historical time is modified.

9. An automatic rating system for the center looseness of continuous casting billets, characterized in that: The system comprises: Acquisition module: used for acquiring a low-magnification sample image of a continuous casting billet to be evaluated, wherein the continuous casting billet is subjected to hot acid corrosion treatment; The first recognition module is used to use the trained segmentation model 1 to perform foreground and background segmentation on the low-magnification sample image to obtain a continuous casting billet image and determine the pixel area S1 of the continuous casting billet; The second recognition module is used to determine whether the continuous casting billet image has central looseness by using the trained segmentation recognition model 2, and if so, segment the central looseness to obtain a central looseness image, and determine the number N of effective central looseness and the pixel area S2 of the loose group region; A data statistics module, for determining a proportion P of a central loose group area based on the pixel area S1 of the continuous casting billet and the pixel area S2 of the loose group area; The rating module is used to comprehensively rate the low center porosity of the continuous casting billet based on the industry standard and the number N of effective center porosity and the area proportion P of the center porosity group area.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method for automatically rating the central looseness of a continuous casting billet as claimed in any one of claims 1 to 8 are implemented.

Images