Method and apparatus for identifying slag removal information in furnace protection from slag splashing, and device and storage medium

Abstract

Provided in the present invention are a method and apparatus for identifying slag removal information in furnace protection from slag splashing, and a device and a storage medium. The method comprises: acquiring an initial image and an image to be identified; on the basis of the labeled initial image, training an initial network model, so as to obtain a slag removal identification model; inputting the image to be identified into the slag removal identification model, so as to obtain an output result of the slag removal identification model, wherein the output result comprises a plurality of different labels; assigning different pixel values to image regions where the different labels in the output result are located, so as to obtain a target slag removal image; and identifying a target slag removal region from the target slag removal image, so as to obtain slag removal information of the image to be identified. An image identification model is constructed by means of an image, so as to perform intelligent identification on a slag removal image to be identified, such that slag removal particles in a converter are identified, thus increasing the accuracy of identification, reducing manpower input, and providing a reliable data source for other instances of intelligent processing.

Description

A method, device, equipment and storage medium for identifying slag splashing and slag jumping information of furnace protection Technical Field

[0001] The present application relates to the technical field of steel smelting, and in particular to a method, device, equipment and storage medium for identifying slag splashing, furnace protection and slag jumping information. Background Art

[0002] In the steelmaking industry, furnace life is a comprehensive technical and economic indicator of converter steelmaking. Mechanical scouring and chemical attack on the furnace lining by high-temperature, highly oxidizing slag are the primary causes of lining erosion. Slag splashing is a commonly used method to improve furnace life. During this process, a high-speed nitrogen jet is injected through a top-blown oxygen lance to impact the high-melting-point slag remaining in the molten pool, causing the slag to splash and coat the entire surface of the converter lining, forming a protective slag layer. In actual production, slag splashing must be monitored in real time during the nitrogen blowing process to adjust the lance height and blowing speed.

[0003] Currently, in the steel smelting industry, the monitoring of slag jumping during slag splashing and furnace protection still relies on manual work, with relatively low accuracy and efficiency. It is impossible to quickly and accurately determine the number, size and location of slag jumping, and it is unable to provide data support for other subsequent intelligent processing.

[0004] Summary of the Invention

[0005] In view of the shortcomings of the existing technology mentioned above, the present invention provides a method, device, equipment and storage medium for identifying slag jumping information in slag splashing furnace protection to solve the technical problems that the monitoring of slag jumping in the current slag splashing furnace protection relies on manual labor, has low accuracy and efficiency, cannot quickly and accurately determine the number, size and position of slag jumping, and cannot provide data support for subsequent other intelligent processing.

[0006] The present invention proposes a method for identifying slag jumping information of slag splashing furnace protection, the method comprising: acquiring an initial image and an image to be identified; training an initial network model based on the labeled initial image to obtain a slag jumping identification model; inputting the image to be identified into the slag jumping identification model to obtain an output result of the slag jumping identification model, the output result comprising a plurality of different labels; assigning different pixel values ​​to image areas where different labels are located in the output result to obtain a target slag jumping image; identifying a target slag jumping area in the target slag jumping image to obtain the slag jumping information of the image to be identified.

[0007] In one embodiment of the present application, an initial model is trained based on an annotated initial image to obtain a skipped slag recognition model, including: obtaining skipped slag information based on the skipped slag image, and obtaining background information based on the non-skipped slag image, the initial image including the skipped slag image and the non-skipped slag image; cropping the initial image to obtain a sample image, generating a first label based on the skipped slag information, generating a second label based on the background information, and annotating the sample image based on the first label and the second label; obtaining a training sample data set based on the annotated sample image, and training the initial network model based on the training sample data set to obtain a skipped slag recognition model.

[0008] In one embodiment of the present application, cropping the initial image to obtain a sample image includes: obtaining preset cropping information and size information of the initial image, the size information including an initial width and an initial height; segmenting the initial image based on the preset cropping information, the initial width, and the initial height to obtain multiple cropped images, and determining the cropped images as sample images.

[0009] In one embodiment of the present application, the sample image is labeled based on the skipping slag information and the background information, including: identifying image information of the sample image, the image information including skipping slag information and background information; determining the area where the skipping slag information is located as a skipping slag block, and determining the area where the background information is located as a background block; labeling the skipping slag block based on a preset first label, and labeling the background area based on a preset second label.

[0010] In one embodiment of the present application, a training sample data set is obtained based on annotated sample images, including: performing data transformation on the annotated sample images to obtain virtual sample images, wherein the data transformation includes at least one transformation method selected from flipping transformation, scaling transformation, translation transformation, contrast transformation, noise perturbation, and Gaussian transformation; and determining the annotated sample images and the virtual sample images as training sample data sets.

[0011] In one embodiment of the present application, different pixel values ​​are assigned to image areas where different labels are located in the output result to obtain a target skipping slag image, including: receiving the output image of the skipping slag recognition model, the output image including a preset first label and a preset second label; assigning a first pixel value to the area where the first label is located, and assigning a second pixel value to the area where the second label is located to obtain a target skipping slag image.

[0012] In one embodiment of the present application, identifying the target slag area in the target slag image includes: determining any slag block in the target slag area as the target slag area, obtaining the individual area of ​​the target slag block, and the target slag area includes at least one slag block; comparing the individual area with a preset area threshold, and when the individual area is greater than the preset minimum area and less than the preset maximum area, determining the target slag area as a legal slag block; traversing each slag block in the target slag area to obtain all legal slag blocks, and determining the area where all the legal slag blocks are located as the target slag area of ​​the image to be identified.

[0013] In one embodiment of the present application, after determining the area where all the legal skipping blocks are located as the target skipping area of ​​the image to be identified, it also includes: calculating the sum of the areas of all the legal skipping blocks based on the individual areas of each skipping block, and determining the sum of the areas as the skipping area of ​​the image to be identified; counting the number of the legal skipping blocks to obtain the number of skipping blocks in the image to be identified; recording the position information of each skipping block to obtain the skipping distribution of the image to be identified; determining the skipping area, the number of skipping blocks, and the skipping distribution information as the skipping information of the image to be identified.

[0014] The present application provides a slag splashing furnace protection slag jumping identification device, which includes: an information acquisition module for acquiring an initial image and an image to be identified; a model construction module for training the initial model based on the labeled initial image to obtain a slag jumping identification model; an image recognition module for inputting the image to be identified into the slag jumping identification model to obtain an output result of the slag jumping identification model, wherein the output result includes multiple different labels; an image analysis module for assigning different pixel values ​​to image areas where different labels are located in the output result to obtain a target slag jumping image; a slag jumping information determination module for identifying a target slag jumping area in the target slag jumping image to obtain the slag jumping information of the image to be identified.

[0015] In one embodiment of the present application, the model construction module includes: an image information recognition submodule, which is used to obtain skipping information based on the skipping image and obtain background information based on the non-skipped image, and the initial image includes the skipping image and the non-skipped image; an image cropping submodule, which is used to crop the initial image to obtain a sample image; an image annotation submodule, which is used to generate a first label based on the skipping information, generate a second label based on the background information, and annotate the sample image based on the first label and the second label; a model training submodule, which is used to obtain a training sample data set based on the annotated sample image, and train the initial network model based on the training sample data set to obtain a skipping recognition model.

[0016] In one embodiment of the present application, a skipping slag information determination module includes: an image receiving submodule for receiving the output image of the skipping slag recognition model, the output image including a preset first label and a preset second label; an image parsing submodule for assigning a first pixel value to the area where the first label is located, and assigning a second pixel value to the area where the second label is located to obtain a target skipping slag image; a skipping slag area determination submodule for calculating the area of ​​each skipping slag block to obtain a legal skipping slag block, and determining the target skipping slag area of ​​the image to be identified based on the legal skipping slag block; a skipping slag information collection submodule for calculating the skipping slag area of ​​the target skipping slag area, counting the number of legal skipping slag blocks, and recording the skipping slag distribution of multiple skipping slag blocks.

[0017] The present application proposes 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 slag splashing furnace protection and slag jumping information identification method as described above.

[0018] The present application proposes a computer-readable storage medium, characterized in that a computer program is stored thereon. When the computer program is executed by a computer processor, the computer is enabled to execute the above-mentioned method for identifying slag splashing and furnace protection slag jumping information.

[0019] Beneficial effects of the present invention: The present invention provides a method, device, equipment and storage medium for identifying slag jumping information of slag splashing furnace protection. The method obtains a slag jumping image, a non-slag jumping image, and an image to be processed, obtains a background image of the initial image based on the non-slag jumping image, determines the part of the slag jumping image that is the same as the background image as the background information, determines the part of the slag jumping image that is different from the background image as the slag jumping information, and annotates the slag jumping image based on the background information and the slag jumping information to obtain a training sample data set, and trains an initial network model based on the training sample data set to obtain a slag jumping recognition model, and inputs the image to be processed into the slag jumping recognition model to identify the slag jumping state of the image to be processed; constructs an image recognition model through the image, intelligently identifies the slag jumping image to be identified, realizes the identification of converter slag jumping particles, improves its identification accuracy, reduces manpower input, and provides a reliable data source for other intelligent processing.

[0020] 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

[0021] 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, explaining the principles of the present application. Obviously, the drawings described below are only some embodiments of the present application, and it is possible for a person of ordinary skill in the art to derive other drawings based on these drawings without inventive effort. In the drawings:

[0022] FIG1 is a schematic diagram of an implementation environment of a method for identifying slag splashing and slag jumping information for furnace protection according to an exemplary embodiment of the present application;

[0023] FIG2 is a flow chart of a method for identifying slag splashing and slag jumping information in an exemplary embodiment of the present application;

[0024] FIG3 is an example of an original image of slag jumping in a method for identifying slag jumping information during slag splashing and furnace protection, shown in an exemplary embodiment of the present application;

[0025] FIG4 is an example of annotated image of an original image in a method for identifying slag splashing and slag jumping information of a furnace protection according to an exemplary embodiment of the present application;

[0026] FIG5 is an example of recognition and segmentation effects in a method for identifying slag splashing and furnace protection slag jumping information, shown in an exemplary embodiment of the present application;

[0027] FIG6 is a block diagram of a slag jumping identification device for slag splashing and furnace protection according to an exemplary embodiment of the present application;

[0028] FIG7 shows a schematic structural diagram of a computer system suitable for implementing an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0029] 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.

[0030] 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.

[0031] 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.

[0032] First, it's important to note that UNet is a variant of FCN. It's arguably the most commonly used and simplest segmentation model. It's simple, efficient, easy to understand, and build, and can be trained from small datasets. FCN (Fully Convolutional Networks for Semantic Segmentation) is a fully convolutional network for semantic segmentation. As the name suggests, the original fully connected layers in the network are replaced by convolutional layers.

[0033] The Transformer is an end-to-end NLP model. It abandons the traditional RNN sequential structure and instead employs a self-attention mechanism to enable parallel training and capture global information. Transformer models are typically trained on large text corpora and then fine-tuned on smaller, specific tasks to achieve optimal computational efficiency and accuracy. The Vision Transformer can be considered a graphical version of the Transformer, directly migrating the standard Transformer model to the image domain with minimal modification.

[0034] FIG1 is a schematic diagram of the implementation environment of the slag splashing furnace protection slag jumping information identification method shown in an exemplary embodiment of the present application. As shown in FIG1 , the implementation environment of the slag splashing furnace protection slag jumping information identification method includes an image acquisition device 101 and a computer device 102, wherein the image acquisition device 101 is used to acquire the original slag jumping image, and the image acquisition device can be any device that can be used to acquire image information, such as a camera or a video camera, and the present application does not impose any restrictions on it. The computer device 102 is specifically configured with a corresponding network algorithm and preset training rules, which are used to train the slag jumping recognition model based on the original slag jumping image acquired by the image acquisition device 101 to obtain a target slag jumping recognition model, and to identify the slag jumping state of the image to be identified. In addition, the computer device 102 can be at least one of a desktop graphics processing unit (GPU) computer, a GPU computing cluster, a neural network computer, etc., or an intelligent processor integrated in the current vehicle, and the present application does not impose any restrictions on this.

[0035] FIG2 is a flow chart of a method for identifying slag splashing and slag jumping information for furnace protection, shown in an exemplary embodiment of the present application.

[0036] As shown in FIG2 , in an exemplary embodiment, the method for identifying slag splashing and slag jumping information for furnace protection includes at least steps S210 to S240, which are described in detail as follows:

[0037] Step S210: obtaining an initial image and an image to be recognized.

[0038] Figure 3 is an example of the original image of slag jumping in the method for identifying slag jumping information for slag splashing and furnace protection shown in an exemplary embodiment of the present application; as shown in Figure 3, there are white bright spots in the image, which are the slag jumping, and the remaining black parts are the background area.

[0039] In one embodiment of the present application, an infrared thermal imaging sensor is used as an example of an image acquisition device. Because the temperature at the slag jumping point is significantly higher than that of the background area, points above or equal to a preset temperature are identified as slag jumping points and displayed in white, while the remaining areas below the preset temperature are identified as background areas and displayed in black. This results in the original image of the slag jumping, or initial image, as shown in Figure 3.

[0040] Step S220: training the initial network model based on the annotated initial image to obtain a skipped slag recognition model.

[0041] In one embodiment of the present application, an initial model is trained based on an annotated initial image to obtain a skip residue recognition model, including: obtaining skip residue information based on the skip residue image, and obtaining background information based on the non-skipped residue image, the initial image including the skip residue image and the non-skipped residue image; cropping the initial image to obtain a sample image, generating a first label based on the skip residue information, generating a second label based on the background information, and annotating the sample image based on the first label and the second label; obtaining a training sample data set based on the annotated sample image, and training the initial network model based on the training sample data set to obtain a skip residue recognition model.

[0042] In one embodiment of the present application, after obtaining the initial image, based on prior knowledge, it can be identified whether the initial image contains a slag jump area. If the initial image contains a slag jump area, the image information of the slag jump area in the initial image is determined as the slag jump information; otherwise, the image information of the non-slag jump area and the image information of the initial image that does not contain the slag jump information are determined as background information.

[0043] In one embodiment of the present application, cropping an initial image to obtain a sample image includes: obtaining preset cropping information and size information of the initial image, the size information including an initial width and an initial height; segmenting the initial image based on the preset cropping information, the initial width, and the initial height to obtain multiple cropped images, and determining the cropped images as sample images.

[0044] In a specific embodiment of the present application, the resolution of the original image is H×W, and the preset cropping information is that the height of the cropped thumbnail is v and the width of the thumbnail is h. Based on the pixel information of the original image, it can be known that the image height of the original image is H, that is, the initial height of the initial image is H, and the image width of the original image is W, that is, the initial width of the initial image is W. Based on its height value H, it is cropped into v parts in the vertical direction, and based on its width value W, it is divided into h parts in the horizontal direction. Based on this, the original image can be cropped into v×h thumbnails, and the cropped thumbnails are stored as sample images in the computer memory. The number of thumbnails is calculated as follows:

[0045] Where X is the number of small images, H is the height of the original image, v is the height of the small image, W is the width of the original image, and h is the width of the small image.

[0046] It should be noted that by cropping the initial image into sample images of smaller size, the amount of computation in the subsequent model training process can be greatly reduced, thereby effectively reducing the memory requirements when training the neural network.

[0047] In one embodiment of the present application, a sample image is labeled based on the skipping slag information and the background information, including: identifying image information of the sample image, the image information including the skipping slag information and the background information; determining the area where the skipping slag information is located as a skipping slag block, and determining the area where the background information is located as a background block; labeling the skipping slag block based on a preset first label, and labeling the background area based on a preset second label.

[0048] It should be noted that after obtaining the above-mentioned initial image, in order to facilitate the subsequent annotation of the image, the initial image needs to be further identified and judged. Its main purpose is to distinguish between the slag jumping area and the non-slag jumping area in the image. Therefore, different labels can be assigned to the slag jumping area image and the non-slag jumping area image respectively, so as to distinguish the attributes of the image area as the slag jumping area or the non-slag jumping area based on the different pixel values. It should be noted that the above-mentioned exemplary description of assigning different labels to images is only an exemplary description method for distinguishing the slag jumping area, and does not impose any restrictions on the specific embodiment and implementation of this application.

[0049] In one embodiment of the present application, (H, W, ) represents a labeled image, where H represents the image height (pixels), W represents the image width (pixels), Represents the label value. The label values ​​of the slag particles and the background are different, that is, the label values ​​of the slag area and the Beijing area in the initial image are different. The initial image is modified to have two channels, namely channel 0 and channel 1. If the pixel is assigned to the slag particles, the value of channel 0 is 1 and the value of channel 1 is 0. If the pixel is assigned to the background, the value of channel 0 is 0 and the value of channel 1 is 1. Therefore, the modified image can be represented as (H, W, 2).

[0050] In a specific embodiment of the present application, a training sample data set is obtained through the above steps, and the pixel values ​​representing the background in the image in the training sample data set are modified to (0, 1), and the pixel values ​​representing the skipped slag particles in the image are modified to (1, 0). Then, the images annotated with this standard are input into the model for training according to the preset training batch number. The specific representation is as follows:

[0051] Among them, pixevalue is the displayed pixel value, (1,0) represents the slag area, and (0,1) represents the background area.

[0052] In a specific embodiment of the present application, a training sample data set is obtained based on the labeled sample image, including: performing data transformation on the labeled sample image to obtain a virtual sample image, the data transformation including at least one transformation method selected from flipping transformation, scaling transformation, translation transformation, contrast transformation, noise perturbation, and Gaussian transformation; and determining the labeled sample image and the virtual sample image as the training sample data set.

[0053] In a specific embodiment of the present application, the labeled sample images are transformed to obtain a training sample dataset, i.e., data augmentation is performed on the labeled dataset. Transformations include, but are not limited to, affine transformation, elastic transformation, horizontal flipping, vertical flipping, sharpening, perspective transformation, piecewise affine, and random pixel dropout. Any of the aforementioned processing methods can be selected to transform the training images, or multiple data processing methods can be randomly selected and randomly combined to process the training images.

[0054] In a specific embodiment of the present application, the labeled sample images are transformed using two data transformation methods: horizontal flipping and vertical flipping. For example, a training sample dataset is obtained. First, the labeled sample images are horizontally flipped to obtain a first dataset. Then, all data in the first dataset are vertically flipped to obtain a second dataset. Finally, the labeled sample images, images in the first dataset, and images in the second dataset are combined to construct a sample dataset.

[0055] It should be noted that there are many ways to obtain new data related to the sample data based on data transformation. The various data transformation methods proposed above are only for illustrative purposes. In the actual data processing process, one or more of the above data transformation methods can be used to process the sample data, or a combination of multiple data processing methods can be selected to transform the sample data. This application does not impose any restrictions on the methods of data transformation during the data processing process. In addition, processing the sample data based on the data transformation methods proposed above can achieve the purpose of data enhancement and effectively increase the total number of training samples in the training data set.

[0056] In addition, it should be noted that the initial network model used to train and generate the skip slag recognition model includes but is not limited to the UNet series, Vision Transformer (ViT) and other neural network models for image segmentation, and can be replaced by other image segmentation networks. This application does not impose any restrictions on the category of the initial network model used to train and generate the skip slag recognition model.

[0057] FIG4 is an example of annotated image of an original image in a method for identifying slag splashing and furnace protection slag jumping information, shown in an exemplary embodiment of the present application.

[0058] In one embodiment of the present application, when data annotation is performed on the slag images used to train the initial network model, images with different degrees of slag and images without slag are selected, and image pixels belonging to slag and background in the training images are annotated with two different labels respectively. The annotated images are stored in a computer memory, and the annotated images are shown in Figure 4.

[0059] Step S230: input the image to be identified into the skip slag identification model to obtain the output result of the skip slag identification model, and the output result includes multiple different labels.

[0060] In one embodiment of the present application, the output result of the slag skipping recognition model includes a first label and a second label, wherein the area where the second label is located is a large block area, and the image area where the second label is located is a scattered block interspersed in the first label.

[0061] Step S240: assign different pixel values ​​to image regions with different labels in the output result to obtain a target slag skipping image.

[0062] In one embodiment of the present application, different pixel values ​​are assigned to image areas where different labels are located in the output result to obtain a target skipping slag image, including: receiving an output image of a skipping slag recognition model, the output image including a preset first label and a preset second label; assigning a first pixel value to the area where the first label is located, and assigning a second pixel value to the area where the second label is located to obtain a target skipping slag image.

[0063] In a specific embodiment of the present application, the image of the skip slag area, i.e., the preset first label area, is assigned a value of 255, and the image of the background area, i.e., the preset second label area, is assigned a value of 0. First, the output image, i.e., the output data, of the skip slag recognition model is obtained, and then the output data is decoded to obtain a grayscale image, which is represented as follows:

[0064] Among them, pixevalue is the displayed pixel value, 255 is the value assigned to the slag particle area in the image, and 0 is the value assigned to the background area in the image.

[0065] In one embodiment of the present application, identifying a target slag region in a target slag image includes: determining any slag block in the target slag region as a target slag region, obtaining a single area of ​​the target slag block, and the target slag region includes at least one slag block; comparing the single area with a preset area threshold, and when the single area is greater than a preset minimum area and less than a preset maximum area, determining the target slag region as a legal slag block; traversing each slag block in the target slag region to obtain all legal slag blocks, and determining the area where all legal slag blocks are located as the target slag region of the image to be identified.

[0066] It should be noted that, because slag skipping has stochasticity, therefore, the slag skipping zone in its image is normally formed jointly by a plurality of disconnected slag skipping blocks.Therefore, when calculating the slag skipping zone area, need to determine the area of ​​each slag skipping block respectively, and obtain overall area based on the area of ​​each little slag skipping block, i.e. the amount slag skipping zone area in this image.In addition, because the area of ​​slag skipping block has uncertainty, therefore for ease of calculation, also be in order to more meet the needs of actual production life, can the slag skipping block that area is too large or area is too small be rejected, with the slag skipping block that obtains each area moderate after screening as the component of slag skipping zone in this image.

[0067] In one embodiment of the present application, the areas identified as slag jumps in the image are screened to remove areas that are too large or too small, to prevent possible misidentified areas from interfering with the subsequent slag jump count. Two thresholds, high bond and low bond, are set to screen the initial slag jump blocks. It should be noted that the threshold should be modified according to the specific production situation. The present application does not impose any restrictions on the specific threshold. The screening steps are as follows:

[0068] First, find the outline of all areas with pixel values ​​of 255 in the image, that is, the outline of the slag block;

[0069] Then, the area within the outline of all the slag-jumping blocks in the image is calculated respectively, that is, the individual area of ​​each slag-jumping block;

[0070] Then, the obtained monomer area is compared with the preset oversized area and undersized area, and the contour data with an area smaller than lowbond and larger than highbond are deleted;

[0071] Finally, after screening the individual areas of each slag skipping block, each slag skipping block that meets the requirements is determined as the target slag skipping block, the sum of the areas of each target slag skipping block is calculated to obtain the slag skipping area of ​​the image to be identified, and the relevant data of each target slag skipping block is statistically analyzed.

[0072] Step S250 , identifying a target slag skipping area in the target slag skipping image to obtain slag skipping information of the image to be identified.

[0073] In one embodiment of the present application, after determining the area where all legal skipping blocks are located as the target skipping area of ​​the image to be identified, it also includes: calculating the sum of the areas of all legal skipping blocks based on the individual areas of each skipping block, and determining the sum of the areas as the skipping area of ​​the image to be identified; counting the number of legal skipping blocks to obtain the number of skipping blocks in the image to be identified; recording the position information of each skipping block to obtain the skipping distribution of the image to be identified; and determining the skipping area, the number of skipping blocks, and the skipping distribution information as the skipping information of the image to be identified.

[0074] In a specific embodiment of the present application, relevant data of each target slag jumping block are counted to obtain the area size, number of target slag jumping areas, and location information of each target slag jumping area in the image to be identified, and a slag jumping record table is generated based on the above relevant information.

[0075] In addition, it should be noted that the above embodiment is based on the skipping slag information obtained after screening the individual areas of each skipping slag block. In actual production life, an initial skipping slag information record table that is the same as the above skipping slag record table can also be obtained based on the information of each unscreened skipping slag block. Therefore, based on actual needs, it is possible to choose to establish a skipping slag record table based on the screened legal skipping slag blocks to obtain the skipping slag information of the image to be identified; or to establish an initial skipping slag record table based on the information of all unscreened skipping slag blocks. This application does not impose any restrictions on whether the skipping slag blocks in the skipping slag information table are screened or not, or on the categories of the skipping slag blocks in the skipping slag information table.

[0076] In one embodiment of the present application, after generating the slag skipping record table, the method further includes: sending the slag skipping record table to an output end to display the slag skipping status of the image to be identified.

[0077] In one embodiment of the present application, a real-time slag jumping image of a converter being generated is used as the image to be identified. After the slag jumping status and the slag jumping record table of the image to be identified are obtained through the above steps, the data of the slag jumping record table is sent to the terminal. After receiving the data of the slag jumping record table, the terminal generates a corresponding chart based on the relevant data, and outputs the relevant chart in a visual form through the terminal's dashboard, so that relevant staff can more intuitively determine the slag jumping situation of the converter and can take corresponding measures in a timely manner based on its specific slag jumping situation.

[0078] In a specific embodiment of the present application, the mobile terminal can generate a real-time slag skipping map based on the location information of each target slag skipping block in the above-mentioned slag skipping record table data to represent the distribution status of each slag skipping block.

[0079] It should be noted that the above-mentioned terminal is a device with a visual dashboard, which can be a smart dashboard, a tablet computer, or any other device that can be used to receive data and generate and display charts based on the received data. This application does not impose any restrictions on the terminal.

[0080] Figure 5 shows an example of the recognition and segmentation effect of a method for identifying slag splashing and slag jumping information, as shown in an exemplary embodiment of the present application. As shown in Figure 5, the image includes bright blocks of color and large black areas. The bright blocks of color are the slag jumping areas, and the black areas are the background areas.

[0081] In one embodiment of the present application, a skip slag recognition model is first constructed based on a large amount of sample data, and then the target image to be identified is input into the skip slag recognition model to obtain a recognition result; then, the recognition result is decoded, that is, different pixel values ​​are assigned to the skip slag area and the non-skip slag area, respectively, to obtain a skip slag image with bright blocks; finally, the area value of each skip slag area in the skip slag image is calculated, and the skip slag blocks whose area exceeds the preset range are eliminated to obtain the final skip slag image as shown in Figure 5.

[0082] Figure 6 is a block diagram of a slag splash protection and slag jump identification device, shown in an exemplary embodiment of the present application. This device can be applied to the implementation environment shown in Figure 1. This device can also be applied to other exemplary implementation environments and specifically configured in other devices. This embodiment does not limit the implementation environment in which this device is applicable.

[0083] As shown in FIG6 , the exemplary slag splashing furnace protection slag jumping identification device includes: an information acquisition module 610 , a model construction module 620 , an image recognition module 630 , an image analysis module 640 , and a slag jumping information determination module 650 .

[0084] Among them, the information acquisition module 610 is used to obtain the initial image and the image to be identified; the model construction module 620 is used to train the initial model based on the labeled initial image to obtain a skipping slag recognition model; the image recognition module 630 is used to input the image to be identified into the skipping slag recognition model to obtain the output result of the skipping slag recognition model, and the output result includes multiple different labels; the image analysis module 640 is used to assign different pixel values ​​to the image areas where different labels are located in the output result to obtain the target skipping slag image; the skipping slag information determination module 650 is used to identify the target skipping slag area in the target skipping slag image to obtain the skipping slag information of the image to be identified.

[0085] In addition, the model construction module includes: an image information recognition sub-module, which is used to obtain skipping information based on the skipping image and obtain background information based on the non-skipped image, and the initial image includes the skipping image and the non-skipped image; an image cropping sub-module, which is used to crop the initial image to obtain a sample image; an image annotation sub-module, which is used to generate a first label based on the skipping information, generate a second label based on the background information, and annotate the sample image based on the first label and the second label; a model training sub-module, which is used to obtain a training sample data set based on the annotated sample image, and train the initial network model based on the training sample data set to obtain a skipping recognition model. The skipping slag information determination module includes: an image receiving submodule for receiving the output image of the skipping slag recognition model, the output image including a preset first label and a preset second label; an image parsing submodule for assigning a first pixel value to the area where the first label is located, and assigning a second pixel value to the area where the second label is located, so as to obtain a target skipping slag image; a skipping slag area determination submodule for calculating the area of ​​each skipping slag block to obtain a legal skipping slag block, and determining the target skipping slag area of ​​the image to be identified based on the legal skipping slag block; a skipping slag information collection submodule for calculating the skipping slag area of ​​the target skipping slag area, counting the number of legal skipping slag blocks, and recording the skipping slag distribution of multiple skipping slag blocks.

[0086] It should be noted that the slag-splattering furnace protection and slag jumping identification device provided in the above embodiment and the slag-splattering furnace protection and slag jumping information identification method provided in the above embodiment belong to the same concept, wherein the specific manner in which each module and unit performs the operation has been described in detail in the method embodiment and will not be repeated here. In actual application, the slag-splattering furnace protection and slag jumping identification device provided in the above embodiment can allocate the above functions to different functional modules as needed, 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.

[0087] 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 slag splashing furnace protection and slag jumping information identification method provided in the above-mentioned embodiments.

[0088] Figure 7 shows a schematic diagram of the structure of a computer system of an electronic device suitable for implementing embodiments of the present application. It should be noted that the computer system 700 of the electronic device shown in Figure 7 is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0089] As shown in Figure 7, computer system 700 includes a central processing unit (CPU) 701, which can perform various appropriate actions and processes according to the program stored in read-only memory (ROM) 702 or the program loaded from storage portion 708 into random access memory (RAM) 703, such as executing the method described in the above embodiment. Various programs and data required for system operation are also stored in RAM 703. CPU 701, ROM 702 and RAM 703 are connected to each other via bus 704. Input / output (I / O) interface 705 is also connected to bus 704.

[0090] The following components are connected to the I / O interface 705: an input section 706 including a keyboard, a mouse, and the like; an output section 707 including devices such as a cathode ray tube (CRT), a liquid crystal display (LCD), and a speaker; a storage section 708 including a hard disk and the like; and a communication section 709 including a network interface card such as a LAN (Local Area Network) card or a modem. The communication section 709 performs communication processing via a network such as the Internet. A drive 710 is also connected to the I / O interface 705 as needed. Removable media 711, such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, is installed in the drive 710 as needed, so that computer programs read therefrom can be installed into the storage section 708 as needed.

[0091] 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 709, and / or installed from a removable medium 711. When the computer program is executed by the central processing unit (CPU) 701, the various functions defined in the system of the present application are executed.

[0092] It should be noted that the computer-readable medium shown in the embodiments of the present application may be a computer-readable signal medium or a computer-readable storage medium or any combination of the two. The computer-readable storage medium may be, for example, an electrical, magnetic, optical, electromagnetic, infrared or semiconductor system, device or component, or any combination of the above. More specific examples of computer-readable storage media may 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 disc 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 may 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 may 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.

[0093] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. 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 boxes can also occur in an order different from that marked in the accompanying drawings. For example, two boxes shown 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, as well as the combination of boxes in the block diagram or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or can be implemented using a combination of dedicated hardware and computer instructions.

[0094] The units described in the embodiments of the present application may be implemented in software or hardware, and the units described may also be provided in a processor. In some cases, the names of these units do not constitute limitations on the units themselves.

[0095] 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 execute the aforementioned method for identifying slag splashing and slag jumping information. 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.

[0096] 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 slag jumping information during slag splashing and furnace protection provided in each of the above embodiments.

[0097] 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 slag splashing and slag jumping information of furnace protection, characterized in that: The method comprises: Obtaining an initial image and an image to be recognized; The initial network model is trained based on the annotated initial image to obtain a skip slag recognition model; Inputting the image to be identified into the skip slag identification model to obtain an output result of the skip slag identification model, wherein the output result includes a plurality of different labels; Assigning different pixel values ​​to image regions where different labels are located in the output result to obtain a target slag skipping image; The target slag skipping area in the target slag skipping image is identified to obtain the slag skipping information of the image to be identified.

2. The method for identifying slag splashing and slag jumping information according to claim 1 is characterized in that: The initial model is trained based on the annotated initial image to obtain a skip slag recognition model, including: Obtaining skipped slag information based on the skipped slag image, and obtaining background information based on the non-skipped slag image, wherein the initial image includes the skipped slag image and the non-skipped slag image; Cropping the initial image to obtain a sample image, generating a first label based on the skipping information, generating a second label based on the background information, and labeling the sample image based on the first label and the second label; A training sample data set is obtained based on the labeled sample images, and the initial network model is trained based on the training sample data set to obtain a skip slag recognition model.

3. The method for identifying slag splashing and slag jumping information of the furnace protection according to claim 2 is characterized in that: Cropping the initial image to obtain a sample image, comprising: Obtaining preset cropping information and size information of the initial image, wherein the size information includes an initial width and an initial height; The initial image is segmented based on the preset cropping information, the initial width, and the initial height to obtain a plurality of cropped images, and the cropped images are determined as sample images.

4. The method for identifying slag splashing and slag jumping information of the furnace protection according to claim 2 is characterized in that: The sample image is labeled based on the skipping information and the background information, including: Identify image information of the sample image, wherein the image information includes skipping information and background information; Determine the area where the skip slag information is located as a skip slag block, and determine the area where the background information is located as a background block; The slag skipping area is labeled based on a preset first label, and the background area is labeled based on a preset second label.

5. The method for identifying slag splashing and slag jumping information of claim 2, characterized in that: The training sample data set is obtained based on the labeled sample images, including: Performing data transformation on the labeled sample image to obtain a virtual sample image, wherein the data transformation includes at least one transformation method selected from the group consisting of flip transformation, scaling transformation, translation transformation, contrast transformation, noise disturbance, and Gaussian transformation; The labeled sample images and the virtual sample images are determined as a training sample data set.

6. The method for identifying slag splashing and slag jumping information according to any one of claims 1 to 5, characterized in that: Assigning different pixel values ​​to image regions where different labels are located in the output result to obtain a target slag skipping image, including: Receiving an output image of the skip slag recognition model, wherein the output image includes a preset first label and a preset second label; The area where the first label is located is assigned a first pixel value, and the area where the second label is located is assigned a second pixel value, so as to obtain a target slag skipping image.

7. The method for identifying slag jumping information of slag splashing furnace protection according to claim 6 is characterized in that: Identifying a target slag skipping area in the target slag skipping image includes: Determine any slag skipping block in the target slag skipping area as the target slag skipping area, and obtain the monomer area of ​​the target slag skipping block, wherein the target slag skipping area includes at least one slag skipping block; Compare the monomer area with a preset area threshold, and when the monomer area is greater than a preset minimum area and less than a preset maximum area, determine the target slag skipping area as a legal slag skipping area; Each skipping block in the target skipping area is traversed to obtain all legal skipping blocks, and the area where all the legal skipping blocks are located is determined as the target skipping area of ​​the image to be identified.

8. The method for identifying slag jumping information of slag splashing furnace protection according to claim 7 is characterized in that: After determining the area where all the legal skipping blocks are located as the target skipping area of ​​the image to be identified, the method further includes: The sum of the areas of all legal slag skipping blocks is calculated based on the individual areas of each slag skipping block, and the sum of the areas is determined as the slag skipping area of ​​the image to be identified; Counting the number of the legal skipped blocks to obtain the number of skipped blocks in the image to be identified; Recording the position information of each slag skipping area to obtain the slag skipping distribution of the image to be identified; The slag skipping area, the number of slag skipping blocks, and the slag skipping distribution information are determined as the slag skipping information of the image to be identified.

9. A slag splashing furnace protection slag jumping identification device, characterized in that: The device comprises: An information acquisition module, used to acquire an initial image and an image to be recognized; A model building module is used to train the initial model based on the annotated initial image to obtain a skip slag recognition model; An image recognition module, used for inputting the image to be recognized into the skip slag recognition model to obtain an output result of the skip slag recognition model, wherein the output result includes a plurality of different labels; An image analysis module, used for assigning different pixel values ​​to image regions where different labels are located in the output result, so as to obtain a target slag skipping image; The skip slag information determination module is used to identify the target skip slag area in the target skip slag image to obtain the skip slag information of the image to be identified.

10. The slag splashing furnace protection slag jumping identification device according to claim 9, characterized in that: The model building module comprises: An image information recognition submodule, used to obtain skipped slag information based on a skipped slag image, and to obtain background information based on a non-skipped slag image, wherein the initial image includes a skipped slag image and a non-skipped slag image; An image cropping submodule, used for cropping the initial image to obtain a sample image; An image annotation submodule, configured to generate a first label based on the skipping information, generate a second label based on the background information, and annotate the sample image based on the first label and the second label; The model training submodule is used to obtain a training sample data set based on the labeled sample images, and to train the initial network model based on the training sample data set to obtain a skip slag recognition model.

11. The slag splashing furnace protection slag jumping identification device according to claim 9, characterized in that: The slag skipping information determination module includes: An image receiving submodule, used for receiving an output image of the skip slag recognition model, wherein the output image includes a preset first label and a preset second label; An image analysis submodule, used for assigning a first pixel value to the area where the first label is located, and assigning a second pixel value to the area where the second label is located, so as to obtain a target slag skipping image; A slag skipping area determination submodule is used to calculate the area of ​​each slag skipping block to obtain a legal slag skipping block, and determine the target slag skipping area of ​​the image to be identified based on the legal slag skipping block; The slag skipping information collection submodule is used to calculate the slag skipping area of ​​the target slag skipping area, count the number of the legal slag skipping blocks, and record the slag skipping distribution of multiple slag skipping blocks.

12. 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 slag splashing and slag jumping information for furnace protection as described in any one of claims 1 to 8.

13. 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 slag jumping information for slag splashing and furnace protection as described in any one of claims 1 to 8.

Images