Melt layered interface online identification and detection method, device, equipment and medium

Abstract

The invention relates to the technical field of image processing, and discloses a melt layered interface online identification and detection method, device and equipment and a medium. The method comprises the following steps: acquiring data of a plurality of images of a gauge rod, and determining the actual height of each pixel point in the data of each image; calibrating the classification category of each piece of image data, constructing a deep learning model, recognizing the image data through the deep learning model to obtain a recognition result, and deploying the deep learning model according to the recognition result of the image data and the corresponding classification category; a plurality of online images of the gauge rod are collected again, target images in the online images are recognized through the deployed deep learning model, and the target images are processed to obtain the pixel range of the upper and lower boundaries of each layered interface of the melt; and calculating the actual height corresponding to each layer of the melt interface according to the actual height of each pixel point and the pixel range of the upper and lower boundaries of each layered interface of the melt. The detection efficiency and accuracy of the melt layered interface are improved, and the labor cost and potential safety hazards are reduced.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to a method, device, equipment and medium for online identification and detection of a melt layering interface. Background Art

[0002] During the copper smelting process, the melt in the smelting furnace is usually divided into three layers, namely the slag layer, the diaphragm layer and the matte layer. The slag layer is located at the top of the melt, mainly composed of oxides and impurities, usually floating on the surface of the melt, and has a lower density; the matte layer is located at the bottom of the melt, is the main copper metal precipitated during the smelting process, and generally has a higher purity; the diaphragm layer is located between the slag layer and the matte layer, is usually thinner, and contains some incompletely reduced metal and alloy components. The interface height of the matte layer, diaphragm layer and slag layer is an important parameter reflecting the state of the smelting furnace. By analyzing the changes in the layered interface height, the effects of different smelting technologies can be evaluated, helping smelting engineers adjust the furnace operating parameters, such as temperature, the amount of flux added, etc., thereby promoting process improvements and optimizing smelting efficiency and copper purity.

[0003] At present, the detection of the interface height of each layer in the smelting furnace at home and abroad still adopts the manual ruler method, that is, inserting the probe rod (usually made of manganese steel) into the smelting furnace, pulling it out after the melt in the furnace leaves a mark on the probe rod, observing the mark color, shape and other characteristics, and measuring the corresponding interface height. Because the smelting furnace is filled with high temperature and toxic gases, the working environment is poor, the labor intensity is high, and the manual ruler measurement itself has observation errors. The differences in the handling habits of different operators will also affect the readings. In addition, when reading the ruler measurement data, the distance to the probe rod is too close, which poses a great safety hazard. Summary of the invention

[0004] In view of this, the purpose of the present invention is to overcome the deficiencies in the prior art and to provide a method, device, equipment and medium for online identification and detection of a melt delamination interface.

[0005] The present invention provides the following technical solutions:

[0006] In a first aspect, an embodiment of the present disclosure provides a method for online identification and detection of a melt delamination interface, the method comprising:

[0007] Collecting multiple image data of the ruler rod, and determining the actual height of each pixel in each of the image data;

[0008] Determine the classification category of each of the image data, construct a deep learning model, identify the image data through the deep learning model to obtain a recognition result, and deploy the deep learning model according to the recognition result of the image data and the corresponding classification category;

[0009] Re-collecting multiple online images of the ruler rod, using the deployed deep learning model to identify the target image in each of the online images, processing the target image, and obtaining the upper and lower pixel ranges of each layered interface of the melt;

[0010] According to the actual height of each pixel point and the upper and lower pixel ranges of each melt layer interface, the actual height corresponding to each layer of the melt interface is calculated.

[0011] Optionally, a ruler is placed on the ruler rod, and determining the actual height of each pixel in each of the image data includes:

[0012] Acquire a first scale on the ruler corresponding to the upper edge of each image data, and acquire a second scale on the ruler corresponding to the lower edge of each image data;

[0013] Taking the absolute value of the difference between the second scale and the first scale as the actual longitudinal distance of each of the image data;

[0014] The number of longitudinal pixels of each of the image data is obtained, and the quotient of the longitudinal actual distance of each of the image data and the corresponding longitudinal pixel number is used as the actual height of each pixel point in each of the image data.

[0015] Optionally, the identifying the image data by using the deep learning model to obtain the identification result includes:

[0016] Dividing all the image data into training data and test data, and using the training data to train the deep learning model;

[0017] The test data is recognized by the trained deep learning model to obtain a recognition result of the test data.

[0018] Optionally, the deploying the deep learning model according to the recognition result of the image data and the corresponding classification category includes:

[0019] Comparing the recognition results of the test data and the corresponding classification categories to obtain multiple comparison results and corresponding test data quantities;

[0020] Calculate the accuracy and recall of the trained deep learning model according to the comparison results and the corresponding test data quantity;

[0021] Determine whether the accuracy is greater than or equal to a preset accuracy threshold, and determine whether the recall is greater than or equal to a preset recall threshold;

[0022] If yes, the deployed deep learning model is generated.

[0023] Optionally, after determining whether the accuracy of the trained deep learning model is greater than a preset accuracy threshold and determining whether the recall rate of the trained deep learning model is greater than a preset recall rate threshold, the method further includes:

[0024] If not, reselect the deep learning model or adjust the training parameters of the deep learning model until the accuracy is greater than or equal to the preset accuracy threshold and the recall is greater than or equal to the preset recall threshold.

[0025] Optionally, the processing of the target image to obtain upper and lower pixel ranges of each layered interface of the melt includes:

[0026] Performing spatial transformation on the target image to obtain a grayscale image, and dividing the grayscale image into a target foreground and a background using an image segmentation algorithm;

[0027] The pixel height range of the target foreground is determined, and the upper and lower pixel ranges of each layered interface of the melt are determined according to the pixel height range of the target foreground.

[0028] Optionally, the calculating the actual height corresponding to each layer of the melt interface according to the actual height of each pixel point and the upper and lower pixel ranges of each layered interface of the melt includes:

[0029] The product of the actual height of each pixel point and the upper and lower pixel ranges of each melt layer interface is used as the actual height corresponding to each layer of the melt interface.

[0030] In a second aspect, an embodiment of the present disclosure provides an online identification and detection device for a melt delamination interface, the device comprising:

[0031] An acquisition module, used for acquiring multiple image data of the ruler rod and determining the actual height of each pixel in each of the image data;

[0032] A construction module is used to calibrate the classification category of each of the image data, construct a deep learning model, identify the image data through the deep learning model, obtain a recognition result, and deploy the deep learning model according to the recognition result of the image data and the corresponding classification category;

[0033] A processing module is used to re-collect multiple online images of the ruler rod, identify the target image in each of the online images using the deployed deep learning model, process the target image, and obtain the upper and lower pixel ranges of each layered interface of the melt;

[0034] The calculation module is used to calculate the actual height corresponding to each layer of the melt interface according to the actual height of each pixel point and the upper and lower pixel ranges of each layered interface of the melt.

[0035] In a third aspect, a computer device is provided in an embodiment of the present disclosure, the computer device comprising a memory and a processor, the memory storing a computer program, and the processor implementing the steps of the method for online identification and detection of the melt delamination interface described in the first aspect when executing the computer program.

[0036] In a fourth aspect, a computer-readable storage medium is provided in an embodiment of the present disclosure, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for online identification and detection of the melt delamination interface described in the first aspect are implemented.

[0037] Beneficial effects of this application:

[0038] The melt layering interface online recognition and detection method provided by the embodiment of the present application includes: collecting multiple image data of the ruler rod, and determining the actual height of each pixel in each of the image data; calibrating the classification category of each of the image data, constructing a deep learning model, identifying the image data through the deep learning model, obtaining the recognition result, and deploying the deep learning model according to the recognition result of the image data and the corresponding classification category; re-collecting multiple online images of the ruler rod, using the deployed deep learning model to identify the target image in each of the online images, processing the target image, and obtaining the upper and lower pixel ranges of each melt layering interface; according to the actual height of each pixel and the upper and lower pixel ranges of each melt layering interface, calculating the actual height corresponding to each layer of the melt interface. The present application uses an area array camera to collect ruler rod image data, and realizes online recognition and detection of melt layering interfaces through a deep learning model, so as to improve the detection efficiency of melt layering interfaces in a standardized manner, reduce the systematic deviation caused by manual observation, and help reduce human operation costs and safety hazards.

[0039] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and understandable, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work. In each of the drawings, similar components are numbered similarly.

[0041] Figure 1A flow chart of an online identification and detection method for a melt delamination interface provided in an embodiment of the present application is shown;

[0042] Figure 2 A schematic diagram of physical mapping of image pixels provided by an embodiment of the present application is shown;

[0043] Figure 3 A schematic diagram of measuring a furnace layer interface provided by an embodiment of the present application is shown;

[0044] Figure 4 A schematic structural diagram of an online melt delamination interface identification and detection device provided in an embodiment of the present application is shown;

[0045] Figure 5 A schematic diagram of the structure of a computer device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0046] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0047] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be a central element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be a central element at the same time. In contrast, when an element is referred to as being "directly on" another element, there is no intermediate element. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.

[0048] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0049] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0050] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein in the specification of the template are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0051] Example 1

[0052] like Figure 1 , which is a flow chart of a method for online identification and detection of a melt delamination interface in an embodiment of the present application, and the method for online identification and detection of a melt delamination interface provided in an embodiment of the present application comprises the following steps:

[0053] Step S110, collecting a plurality of image data of the ruler rod, and determining the actual height of each pixel in each of the image data.

[0054] First of all, it should be noted that when using a measuring rod to measure the height of the melt interface, the measuring rod is generally inserted into the smelting furnace and lifted up after a period of time. Under the action of high temperature, the part of the measuring rod in contact with the melt will usually have a color difference, appearing yellow or white, and the part not in contact with the melt or the measuring rod in a cooled state will usually appear gray. For the matte layer, diaphragm layer and slag layer, the cooling rates within a fixed time are different, and the color changes under different cooling states are different. By investigating the process of manually measuring the melt interface height with a ruler, parameters such as the height of the measuring rod and the height of the smelting furnace body are obtained. By way of example, the following table shows the process of manually measuring the melt layer interface height in a smelter:

[0055] step operate Release Manually insert the measuring rod slowly into the melting furnace Stand Wait for a while, usually 3-5 minutes Lifting rod Slowly lift the measuring rod to a certain height at a certain speed Layered judgment Determine the melt layering position based on the color or adhesion status of the ruler reading Measure melt layer height based on layer position

[0056] Furthermore, the installation position of the area array camera is determined according to the moving range of the on-site ruler rod, the vertical height of the center point of the area array camera from the horizontal ground and the horizontal distance from the ruler rod are obtained, and multiple image data of the ruler rod are collected at a certain time interval. The sample space composed of the multiple image data is A:

[0057] A=(a1,a2,a3,…a n )

[0058] Among them, a i(i=1,2,3…n) is the collected image data.

[0059] This embodiment takes the collection of 1000 images as an example, that is, there are:

[0060] A=(a1,a2,a3,…a 1000 )

[0061] It should be noted that if Figure 2 As shown, a sufficiently long ruler is placed on the ruler rod. Then, it is necessary to record the first scale H1 on the ruler corresponding to the upper edge of the image data collected by the area array camera at this time, and record the second scale H2 on the ruler corresponding to the lower edge of the image data collected by the area array camera at this time, and then calculate the actual longitudinal distance H of the image data collected by the area array camera:

[0062] H=|H1-H2|

[0063] For example, when H1=1500 mm and H2=500 mm, then H=1000 mm.

[0064] Understandably, since the image resolution collected by the area array camera is fixed, the number of vertical pixels pixelH of each image data is obtained, and then the actual height h corresponding to each pixel point in each image data is calculated according to the actual vertical distance H of the image data:

[0065]

[0066] For example, when pixelH=100 and H=1000 mm, h=10 mm.

[0067] The above method realizes the conversion from image data to actual physical size by calculating the actual height of each pixel, which provides the necessary parameters for the final calculation of the actual height of each layer of the melt interface.

[0068] Step S120, calibrate the classification category of each of the image data, construct a deep learning model, identify the image data through the deep learning model to obtain a recognition result, and deploy the deep learning model according to the recognition result of the image data and the corresponding classification category.

[0069] It should be noted that the color of the part of the measuring rod that contacts the melt is affected by the temperature in the smelting furnace and the time it is inserted into the furnace. The temperature mainly affects the time it is inserted into the furnace. Under the same conditions, high temperature makes it take a shorter time for the measuring rod to reach the same adhesion state. If the measuring rod is inserted into the furnace for too short a time, the adhesion state may not be obvious and the stratification state may be difficult to distinguish; if the measuring rod is inserted into the furnace for too long, the hardness of the measuring rod may decrease and bend, affecting the normal measurement results. Therefore, this application proposes the need to first identify whether the measurement state of the measuring rod is normal.

[0070] Understandably, first define the category sample space Y of image data, there is:

[0071] Y=(y1,y2,y3…y p )

[0072] Among them, y i (i=1, 2, 3...p) is the classification category of image data.

[0073] Exemplarily, in this embodiment, the classification categories of image data are 4 categories, namely:

[0074] Y=(y1,y2,y3,y4)

[0075] The meaning of each category is as follows:

[0076] y1: The ruler rod is in normal state and the target image meets the melt layer recognition conditions;

[0077] y2: The ruler rod is in an abnormal state, with no obvious bending or delamination;

[0078] y3: The range of the ruler is abnormal, and the moving speed is too fast to capture the complete image;

[0079] y4: The background of the ruler is blocked, and the camera is blocked by environmental factors or human influence.

[0080] Furthermore, the calibration image dataset A(a1,a2,a3,…a n ) corresponds to the classification category Q, and there exists:

[0081] Q=(q1,q2,q3…q n )

[0082] Among them, q j =y i (i=1,2,3…p,j=1,2,3…n) is the classification category corresponding to the image data.

[0083] For example, the calibration image acquisition data set A (a1, a2, a3, ... a 1000 ) corresponds to the classification category Q = (q1, q2, q3…q1000 ).

[0084] Next, all image data A (a1, a2, a3, ... a n ) is divided into training data and test data, and the sample space B of the training data and the sample space C of the test data are defined. There exists:

[0085] B=(b1,b2,b3,…b k )

[0086] C=(c1,c2,c3…c m )

[0087] Among them, k+m=n, b i (i=1,2,3…k) is the training data, c i (i=1,2,3…m) is the test data.

[0088] Exemplarily, in this embodiment, the ratio of training data to test data is set to 7:3, then:

[0089] B=(b1,b2,b3,…b 700 )

[0090] C=(c1,c2,c3…c 300 )

[0091] Furthermore, a deep learning model M is constructed, and the training data B = (b1, b2, b3, ... b k ) to train the deep learning model. Then the trained deep learning model is used to test the test data C=(c1,c2,c3…c m ) to identify and obtain the identification results of the test data, there are:

[0092] T=(t1,t2,t3…t m )=M(C)

[0093] Among them, t i (i=1,2,3…m) is the trained deep learning model M for the test data c i The recognition result, that is, t i =M(c i ).

[0094] For example, the trained deep learning model is used to train the test data C = (c1, c2, c3 ... c 300 ) is identified by T = (t1, t2, t3…t 300 )=M(C).

[0095] Next, compare the recognition results of the test data T(t1, t2, t3…tm ) and the corresponding classification category Q = (q1, q2, q3…q m ), and obtain multiple comparison results and the corresponding number of test data.

[0096] It should be noted that the comparison results include the following four types: (1) True Positives (TP): refers to the actual classification category is the positive class (i.e., target image), and the recognition result predicted by the model is also the positive class, that is, the model correctly recognizes the sample of the target image; (2) True Negatives (TN): refers to the actual classification category is the negative class (non-target image), and the recognition result predicted by the model is also the negative class, that is, the model correctly recognizes the sample of the non-target image; (3) False Positives (FP): refers to the actual classification category is the negative class (non-target image), but the recognition result predicted by the model is the positive class, that is, the model mistakenly recognizes the sample of the non-target image as the target image; (4) False Negatives (FN): refers to the actual classification category is the positive class (i.e., target image), but the recognition result predicted by the model is the negative class, that is, the model mistakenly recognizes the sample of the target image as the non-target image.

[0097] For example, the recognition results T(t1, t2, t3…t 300 ) and the corresponding classification category Q = (q1, q2, q3…q 300 ), the comparison results obtained in this embodiment and the corresponding test data quantity distribution are as follows:

[0098] TP 132 Actual is positive, model predicts positive TN 147 Actual is negative, simulation predicts negative FP 8 The actual value is negative, but the model predicts it to be positive. FN 13 Actual is positive, model predicts negative

[0099] Furthermore, according to each comparison result and the corresponding number of test data, the accuracy Pre and recall Re of the trained deep learning model are calculated:

[0100]

[0101] Among them, TP+TN+FP+FN=m.

[0102] Further, it is determined whether the preset target conditions are met, that is, it is determined whether the accuracy Pre is greater than or equal to the preset accuracy threshold Pre_val, and it is determined whether the recall rate Re is greater than or equal to the preset recall rate threshold Re_val. If so, that is, Pre≥Pre_val and Re≥Re_val, the trained deep learning model is deployed to generate a deployed deep learning model; if not, that is, Pre≥Pre_val and Re<Re_val, or Pre<Pre_val and Re<Re_val, or Pre<Pre_val and Re≥Re_val, the deep learning model is reselected or the training parameters of the deep learning model are adjusted until the preset target conditions are met, that is, the accuracy is greater than or equal to the preset accuracy threshold and the recall rate is greater than or equal to the preset recall rate threshold.

[0103] The above method can achieve accurate recognition of image data by constructing a deep learning model, training it using training data, and validating the model using test data. At the same time, dividing the image data into different classification categories helps the model to more accurately judge the state of the measuring rod during the recognition process, thereby improving the recognition accuracy. In addition, by comparing the recognition results of the test data with the corresponding classification categories, the accuracy and recall rate of the model can be calculated, and then the model can be optimized to make it more adaptable to the actual application scenario.

[0104] Step S130, re-collect multiple online images of the measuring rod, use the deployed deep learning model to recognize the target images in each of the online images, and process the target images to obtain the pixel range of the upper and lower bounds of each stratified interface of the melt.

[0105] Preferably, re-collect multiple online images of the measuring rod, define the sample space as D, and there is:

[0106] D=(d1,d2,d3…d s )

[0107] where d i (i = 1,2,3…s) is an online image.

[0108] Exemplarily, in this embodiment, 16 online images are re-collected, that is, there is:

[0109] D=(d1,d2,d3…d 16 )

[0110] Then, use the deployed deep learning model to classify and recognize each online image to obtain the classification category y i of each online image, that is, y i =M(d i ).

[0111] Further, determine the classification category y i Is it the target image? Determine the target image in each online image. If the classification category y i If it is not the target image, continue to re-acquire the online image of the ruler rod.

[0112] Exemplarily, in this embodiment, the fifth online image is the target image, that is, y1 = M(d5).

[0113] Understandably, the target image defaults to RGB color channels. i Perform spatial transformation to obtain the grayscale image g. Each pixel corresponds to three channels: R, G, and B. The grayscale value conversion formula is as follows:

[0114] Gray=R×0.299+G×0.587+B×0.114

[0115] It should be noted that the fixed position of the area array camera needs to be fixed in front of the measuring rod at a certain distance to ensure that the movement range of the measuring rod relative to the measuring point can be captured. In the on-site environment, there will inevitably be the influence of environmental factors such as human movement or smoke and dust, which will cause certain interference to the camera sampling, so it is necessary to identify the abnormal background.

[0116] Therefore, the present application further uses the image segmentation algorithm U to divide the grayscale image g into two parts: target foreground g1 and background g2, that is, g1, g2 = U (g). The target foreground is the main target to be identified, that is, the ruler in a layered state, such as Figure 3 As shown in Figure 2, the pixel height range is recorded as: r1-r4, where r1 represents the upper boundary (or starting) pixel position of the uppermost layer (usually the slag layer) in the melt layer interface. Figure 3 In the figure, this position is the boundary between the slag layer and the background or air above; r2 represents the lower boundary (or end) pixel position of the slag layer, which is also the upper boundary pixel position of the next layer (usually the diaphragm layer), and this position is the boundary between the slag layer and the diaphragm layer; r3 represents the lower boundary (or end) pixel position of the diaphragm layer, which is also the upper boundary pixel position of the next layer (usually the matte layer), and this position is the boundary between the diaphragm layer and the matte layer; r4 represents the lower boundary (or final) pixel position of the lowest layer (usually the matte layer) in the melt stratification interface, and this position is the boundary between the matte layer and other materials below or the furnace bottom.

[0117] It should also be noted that, in general, the gauge rod moves in the vertical direction. Due to the influence of manual operation differences and the sampling cycle of the area array camera, if the gauge rod moves too fast, it may be difficult to capture the contact part between the gauge rod and the melt or the capture range is limited during the lifting process. It is necessary to locate and identify the gauge rod during the movement process to ensure the range of the contact part between the gauge rod and the melt. Figure 3 As shown in the figure, the image connected domain is calculated for the pixel height range of the target foreground, and the pixel positions corresponding to the upper and lower boundaries of each layered interface of the melt are obtained, where:

[0118] (1) The pixel range of the slag layer is recorded as: r1-r2;

[0119] (2) The pixel range of the diaphragm layer is recorded as: r2-r3;

[0120] (3) The pixel range of the matte layer is recorded as: r3-r4.

[0121] Exemplarily, in this embodiment, the pixel height range of the target foreground is: r1-r4=100, and the pixel point positions corresponding to the upper and lower boundaries of each layer interface of the melt are: slag layer pixel range r1-r2=20, diaphragm layer pixel range r2-r3=30, matte layer pixel range r3-r4=50.

[0122] The above method uses the deployed deep learning model to identify online images, which can realize real-time monitoring of the state of the ruler rod, so as to obtain the information of each layer interface of the melt in time. At the same time, by classifying and identifying online images, the target image, that is, the image that meets the melt layer recognition conditions, is screened out, which helps to reduce the amount of calculation for subsequent processing and improve processing efficiency. In addition, the image segmentation algorithm is used to divide the target foreground and background, which can eliminate the interference of abnormal background on the recognition result and improve the accuracy of recognition.

[0123] Step S140, calculating the actual height of each layer of the melt interface according to the actual height of each pixel point and the upper and lower pixel ranges of each layered interface of the melt.

[0124] Finally, the actual height h corresponding to the pixel point determined in step S110 is multiplied by the upper and lower pixel ranges of the slag layer, diaphragm layer and matte layer determined in step S130 as the actual height corresponding to each layer of the melt interface, that is:

[0125] (1) The corresponding height of the slag layer is: h×(r1-r2);

[0126] (2) The corresponding height of the diaphragm layer is: h×(r2-r3);

[0127] (3) The corresponding height of the matte layer is: h×(r3-r4).

[0128] For example, in this embodiment, the actual height corresponding to the slag layer is: h×(r1-r2)=10×20=200mm, the actual height corresponding to the diaphragm layer is: h×(r2-r3)=10×30=300mm, and the actual height corresponding to the matte layer is: h×(r3-r4)=10×50=500mm.

[0129] The above method can accurately calculate the actual height of each layer of the melt interface according to the actual height of each pixel point and the upper and lower pixel ranges of each layered interface of the melt, which can provide support for subsequent production decisions, such as adjusting smelting conditions and optimizing process parameters.

[0130] The melt layering interface online recognition and detection method provided in the embodiment of the present application is to collect multiple image data of the ruler rod and determine the actual height of each pixel in each of the image data; calibrate the classification category of each of the image data, build a deep learning model, identify the image data through the deep learning model, obtain the recognition result, and deploy the deep learning model according to the recognition result of the image data and the corresponding classification category; re-collect multiple online images of the ruler rod, use the deployed deep learning model to identify the target image in each of the online images, process the target image, and obtain the upper and lower pixel ranges of each melt layering interface; calculate the actual height corresponding to each layer of the melt interface according to the actual height of each pixel and the upper and lower pixel ranges of each melt layering interface. The present application uses an area array camera to collect ruler rod image data, and realizes online recognition and detection of melt layering interfaces through a deep learning model, so as to improve the detection efficiency of melt layering interfaces in a standardized manner, reduce the systematic deviation caused by manual observation, and help reduce human operation costs and safety hazards.

[0131] Example 2

[0132] like Figure 4 , which is a schematic diagram of the structure of an online melt layering interface identification and detection device 400 in an embodiment of the present application, and the device includes:

[0133] An acquisition module, used for acquiring multiple image data of the ruler rod and determining the actual height of each pixel in each of the image data;

[0134] A construction module is used to calibrate the classification category of each of the image data, construct a deep learning model, identify the image data through the deep learning model, obtain a recognition result, and deploy the deep learning model according to the recognition result of the image data and the corresponding classification category;

[0135] A processing module is used to re-collect multiple online images of the ruler rod, identify the target image in each of the online images using the deployed deep learning model, process the target image, and obtain the upper and lower pixel ranges of each layered interface of the melt;

[0136] The calculation module is used to calculate the actual height corresponding to each layer of the melt interface according to the actual height of each pixel point and the upper and lower pixel ranges of each layered interface of the melt.

[0137] The melt delamination interface online recognition and detection device provided in the embodiment of the present application uses an area array camera to collect image data of a measuring rod, and realizes online recognition, detection and detection of the melt delamination interface through a deep learning model, thereby improving the detection efficiency of the melt delamination interface in a standardized manner, reducing the systematic deviation caused by manual observation, and helping to reduce manual operation costs and safety hazards.

[0138] Example 3

[0139] The present application also provides a computer device. Figure 5 , Figure 5 This is a basic structural block diagram of the computer device in this embodiment.

[0140] The computer device 5 includes a memory 51, a processor 52, and a network interface 53 that are interconnected through a system bus. It should be noted that the figure only shows a computer device 5 with a memory 51, a processor 52, and a network interface 53, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), programmable gate arrays (Field-Programmable Gate Array, FPGA), digital processors (Digital Signal Processor, DSP), embedded devices, etc.

[0141] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.

[0142] The memory 51 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or D slot compatibility test memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 51 can be an internal storage unit of the computer device 5, such as a hard disk or memory of the computer device 5. In other embodiments, the memory 51 can also be an external storage device of the computer device 5, such as a plug-in hard disk equipped on the computer device 5, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. Of course, the memory 51 can also include both the internal storage unit of the computer device 5 and its external storage device. In this embodiment, the memory 51 is generally used to store the operating system and various application software installed on the computer device 5, such as computer-readable instructions of the slot compatibility test method, etc. In addition, the memory 51 can also be used to temporarily store various types of data that have been output or are to be output.

[0143] The processor 52 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other melt layer interface online identification and detection chip in some embodiments. The processor 52 is generally used to control the overall operation of the computer device 5. In this embodiment, the processor 52 is used to run the computer-readable instructions or process data stored in the memory 51, such as computer-readable instructions for running the slot compatibility test method.

[0144] The network interface 53 may include a wireless network interface or a wired network interface. The network interface 53 is generally used to establish a communication connection between the computer device 5 and other electronic devices.

[0145] The computer device provided in this embodiment can execute the above-mentioned melt delamination interface online identification and detection method. Here, the melt delamination interface online identification and detection method can be the melt delamination interface online identification and detection method of each of the above-mentioned embodiments.

[0146] Example 4

[0147] This embodiment further provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the method for online identification and detection of the melt delamination interface in the embodiment are implemented.

[0148] In this embodiment, the computer-readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the computer-readable storage medium can be an internal storage unit of a computer device, such as a hard disk or memory of the computer device. In other embodiments, the computer-readable storage medium can also be an external storage device of a computer device, such as a plug-in hard disk equipped on the computer device, a smart memory card (Smart Media Card, referred to as SMC), a secure digital (Secure Digital, referred to as SD) card, a flash card, etc. Of course, the computer-readable storage medium can also include both the internal storage unit of the computer device and its external storage device. In this embodiment, the computer-readable storage medium is generally used to store an operating system and various application software installed on the computer device. In addition, the computer-readable storage medium can also be used to temporarily store various types of data that have been output or are to be output.

[0149] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device embodiments described above are merely schematic. For example, the flowcharts and structure diagrams in the accompanying drawings show the possible architecture, functions and operations of the devices, methods and computer program products according to multiple embodiments of the present invention. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the module, a program segment or a part of a code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in an alternative implementation, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes 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 structure diagram and / or the flow chart, and the combination of boxes in the structure diagram and / or the flow chart, can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0150] In addition, the functional modules or units in the various embodiments of the present invention may be integrated together to form an independent part, or each module may exist independently, or two or more modules may be integrated to form an independent part.

[0151] If the function is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or partly contributed to the prior art or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a smart phone, a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium can be a non-volatile storage medium or a volatile storage medium. For example, the storage medium can be: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a disk or an optical disk, and other media that can store program codes.

[0152] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for online identification and detection of melt delamination interface, characterized in that: The method comprises: Collecting multiple image data of the ruler rod, and determining the actual height of each pixel in each of the image data; Determine the classification category of each of the image data, construct a deep learning model, identify the image data through the deep learning model to obtain a recognition result, and deploy the deep learning model according to the recognition result of the image data and the corresponding classification category; Re-collecting multiple online images of the ruler rod, using the deployed deep learning model to identify the target image in each of the online images, processing the target image, and obtaining the upper and lower pixel ranges of each layered interface of the melt; According to the actual height of each pixel point and the upper and lower pixel ranges of each melt layer interface, the actual height corresponding to each layer of the melt interface is calculated.

2. The method for online identification and detection of melt delamination interface according to claim 1, characterized in that: A ruler is placed on the ruler rod, and determining the actual height of each pixel in each of the image data includes: Acquire a first scale on the ruler corresponding to the upper edge of each image data, and acquire a second scale on the ruler corresponding to the lower edge of each image data; Taking the absolute value of the difference between the second scale and the first scale as the actual longitudinal distance of each of the image data; The number of longitudinal pixels of each of the image data is obtained, and the quotient of the longitudinal actual distance of each of the image data and the corresponding longitudinal pixel number is used as the actual height of each pixel point in each of the image data.

3. The method for online identification and detection of melt delamination interface according to claim 1, characterized in that: The identifying the image data by using the deep learning model to obtain the identification result includes: Dividing all the image data into training data and test data, and using the training data to train the deep learning model; The test data is recognized by the trained deep learning model to obtain a recognition result of the test data.

4. The method for online identification and detection of melt delamination interface according to claim 3, characterized in that: The deploying the deep learning model according to the recognition result of the image data and the corresponding classification category includes: Comparing the recognition results of the test data and the corresponding classification categories to obtain multiple comparison results and corresponding test data quantities; Calculate the accuracy and recall of the trained deep learning model according to the comparison results and the corresponding test data quantity; Determine whether the accuracy is greater than or equal to a preset accuracy threshold, and determine whether the recall is greater than or equal to a preset recall threshold; If yes, the deployed deep learning model is generated.

5. The method for online identification and detection of melt delamination interface according to claim 4, characterized in that: After determining whether the accuracy of the trained deep learning model is greater than a preset accuracy threshold and determining whether the recall rate of the trained deep learning model is greater than a preset recall rate threshold, the method further includes: If not, reselect the deep learning model or adjust the training parameters of the deep learning model until the accuracy is greater than or equal to the preset accuracy threshold and the recall is greater than or equal to the preset recall threshold.

6. The method for online identification and detection of melt delamination interface according to claim 1, characterized in that: The target image is processed to obtain upper and lower pixel ranges of each melt layer interface, including: Performing spatial transformation on the target image to obtain a grayscale image, and dividing the grayscale image into a target foreground and a background using an image segmentation algorithm; The pixel height range of the target foreground is determined, and the upper and lower pixel ranges of each layered interface of the melt are determined according to the pixel height range of the target foreground.

7. The method for online identification and detection of melt delamination interface according to claim 2, characterized in that: The actual height corresponding to each layer of the melt interface is calculated according to the actual height of each pixel point and the upper and lower pixel ranges of each layered interface of the melt, including: The product of the actual height of each pixel point and the upper and lower pixel ranges of each melt layer interface is used as the actual height corresponding to each layer of the melt interface.

8. A melt layer interface online identification and detection device, characterized in that: The device comprises: An acquisition module, used for acquiring multiple image data of the ruler rod and determining the actual height of each pixel in each of the image data; A construction module is used to calibrate the classification category of each of the image data, construct a deep learning model, identify the image data through the deep learning model, obtain a recognition result, and deploy the deep learning model according to the recognition result of the image data and the corresponding classification category; A processing module is used to re-collect multiple online images of the ruler rod, identify the target image in each of the online images using the deployed deep learning model, process the target image, and obtain the upper and lower pixel ranges of each layered interface of the melt; The calculation module is used to calculate the actual height corresponding to each layer of the melt interface according to the actual height of each pixel point and the upper and lower pixel ranges of each layered interface of the melt.

9. A computer device, characterized in that: The invention comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method for online identification and detection of melt delamination interface according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for online identification and detection of melt delamination interface according to any one of claims 1 to 7 are implemented.

Images