Intelligent identification and performance determination method, system and storage medium for equiaxed crystals in ingots

Abstract

The embodiments of this specification disclose a method, system and storage medium for intelligent identification and performance determination of equiaxed crystals in ingots. Through image processing technology and semantic segmentation technology, intelligent and accurate automatic segmentation of equiaxed crystal areas in equiaxed crystal samples is achieved. Combined with computer graphics measurement technology, automatic measurement of equiaxed crystal areas, automatic calculation of equiaxed crystal ratio and evaluation of equiaxed crystal ratio distribution are achieved. Further, the equiaxed crystal performance of equiaxed crystal samples is determined based on the equiaxed crystal ratio and equiaxed crystal ratio distribution, thereby achieving intelligent determination of the equiaxed crystal performance of the samples, and solving the problems of poor efficiency, low accuracy, poor identity and unscientific nature of existing manual measurement of equiaxed crystals.

Description

Technical Field

[0001] The embodiments of this specification relate to the field of iron and steel metallurgical automation and image processing, and in particular to a method, system and storage medium for intelligent identification and performance determination of equiaxed crystals in ingots. Background Art

[0002] Continuous casting is the main method of steel production in my country. The quality of continuous casting billets directly determines the quality of finished steel products to a large extent. The quality of continuous casting billets is closely related to their solidification structure, which is usually composed of three parts: fine equiaxed crystals on the surface, columnar crystals in the middle, and equiaxed crystals in the center. The solidification structure of continuous casting billets has a great influence on the processing performance and mechanical properties of steel. Except for some special steel types that require columnar crystal structure, most steel types hope to obtain billet structures with large equiaxed crystal areas, and at the same time hope that the equiaxed crystals are small and uniform. The greater the proportion of equiaxed crystals in the billet, the better the comprehensive mechanical properties of the steel. Therefore, the equiaxed crystal rate of the billet is a key factor in measuring the quality of the billet.

[0003] At present, most domestic steel plant technicians analyze the influence of different continuous casting process parameters on the equiaxed crystal rate at the center of the ingot by observing the macrostructure of the ingot, so as to optimize the continuous casting process parameters. Although the existing ingot equiaxed crystal rate detection method can quantitatively measure the ingot structure, the division, measurement and evaluation of the equiaxed crystal area mainly rely on manual work, and there are still the following problems:

[0004] 1. The columnar crystal area of ​​the ingot blends with the central equiaxed crystal area and the edge equiaxed crystal area, and their boundaries are not straight lines. When manually processed, straight lines are often used for division, which results in certain segmentation errors and affects the measurement results of the equiaxed crystal ratio.

[0005] 2. Based on the manual division of regions, the sizes of different grain regions are manually measured using a tape measure. This is limited by the skill level of different measurement personnel and has the disadvantages of low measurement accuracy, low measurement efficiency, and poor consistency of measurement and evaluation results.

[0006] 3. The microstructure and performance of the ingot are not only related to the equiaxed crystal ratio, but also closely related to the equiaxed crystal distribution. At present, the existing manual evaluation of the microstructure and performance of the ingot is mainly based on the equiaxed crystal ratio. There is no corresponding detection method for the equiaxed crystal distribution, and there is a lack of corresponding evaluation methods, which cannot meet the needs of comprehensive on-site evaluation of the microstructure and performance of the ingot. Summary of the invention

[0007] The embodiments of this specification provide a method, system and storage medium for intelligent identification and performance determination of equiaxed crystals in a casting.

[0008] In a first aspect, the embodiments of this specification provide a method for intelligently identifying and determining the performance of equiaxed crystals in a casting, the method comprising:

[0009] Acquire an equiaxed crystal sample image of the equiaxed crystal sample;

[0010] The equiaxed crystal sample image of the equiaxed crystal sample is segmented by a pre-trained semantic segmentation model to obtain the equiaxed crystal region;

[0011] The equiaxed crystal region is measured to obtain the equiaxed crystal ratio and equiaxed crystal ratio distribution;

[0012] The equiaxed crystal properties of the equiaxed crystal sample are determined based on the equiaxed crystal ratio and the equiaxed crystal ratio distribution.

[0013] Optionally, the training method of the semantic segmentation model is:

[0014] Preprocessing the collected equiaxed crystal sample image to obtain a preprocessed image;

[0015] Obtaining an equiaxed crystal sample image training set, wherein the equiaxed crystal sample image training set includes a plurality of unlabeled preprocessed images and a plurality of labeled equiaxed crystal region preprocessed images;

[0016] Training an equiaxed crystal segmentation network model based on the equiaxed crystal sample image training set;

[0017] When the loss function of the equiaxed crystal segmentation network model converges or the number of training times reaches a predetermined value, the training of the equiaxed crystal segmentation network model is terminated to obtain a trained equiaxed crystal segmentation network model.

[0018] Optionally, measuring the equiaxed crystal region to obtain the equiaxed crystal ratio and the equiaxed crystal ratio distribution includes:

[0019] The equiaxed crystal area of ​​the sample and the total area of ​​the sample are obtained; the equiaxed crystal area of ​​the sample is the area of ​​the equiaxed crystal area in the equiaxed crystal sample image, and the total area of ​​the sample is the area occupied by the sample ingot in the equiaxed crystal sample image;

[0020] Based on the area of ​​the equiaxed crystal region and the total area of ​​the sample, obtaining an equiaxed crystal ratio;

[0021] The equiaxed crystal ratio distribution was obtained based on the equiaxed crystal ratio of the ingot sample along the width direction of the ingot.

[0022] Optionally, the equiaxed crystal sample image is preprocessed to obtain a preprocessed image, including:

[0023] The equiaxed crystal sample image is subjected to grayscale transformation, tilt correction, image denoising and image scaling to obtain a preprocessed image.

[0024] In a second aspect, an embodiment of the present invention further provides a system for intelligently identifying and determining the performance of equiaxed crystals in a casting, the system comprising an equiaxed crystal image acquisition module and an equiaxed crystal intelligent detection and grading module; wherein:

[0025] The equiaxed crystal image acquisition module is used to obtain an equiaxed crystal sample image of the equiaxed crystal sample;

[0026] The equiaxed crystal intelligent detection and grading module includes an equiaxed crystal semantic segmentation module, an equiaxed crystal measurement module, a casting blank parameter module, and an equiaxed crystal grading module;

[0027] The equiaxed crystal semantic segmentation module is used to segment the equiaxed crystal sample image of the equiaxed crystal sample by using a pre-trained semantic segmentation model to obtain an equiaxed crystal region;

[0028] The equiaxed crystal measurement module is used to measure the equiaxed crystal area to obtain the equiaxed crystal ratio and equiaxed crystal ratio distribution;

[0029] The casting parameter module is used to store the process parameters of casting production;

[0030] The equiaxed crystal grading module is used to determine the equiaxed crystal performance of the equiaxed crystal sample based on the equiaxed crystal ratio and the equiaxed crystal ratio distribution.

[0031] Optionally, the equiaxed crystal semantic segmentation module is further used for:

[0032] Preprocessing the collected equiaxed crystal sample image to obtain a preprocessed image;

[0033] Obtaining an equiaxed crystal sample image training set, wherein the equiaxed crystal sample image training set includes a plurality of unlabeled preprocessed images and a plurality of labeled equiaxed crystal region preprocessed images;

[0034] Training an equiaxed crystal segmentation network model based on the equiaxed crystal sample image training set;

[0035] When the loss function of the equiaxed crystal segmentation network model converges or the number of training times reaches a predetermined value, the training of the equiaxed crystal segmentation network model is terminated to obtain a trained equiaxed crystal segmentation network model.

[0036] Optionally, the equiaxed crystal measurement module is specifically used for:

[0037] The equiaxed crystal area of ​​the sample and the total area of ​​the sample are obtained; the equiaxed crystal area of ​​the sample is the area of ​​the equiaxed crystal area in the equiaxed crystal sample image, and the total area of ​​the sample is the area occupied by the sample ingot in the equiaxed crystal sample image;

[0038] Based on the area of ​​the equiaxed crystal region and the total area of ​​the sample, obtaining an equiaxed crystal ratio;

[0039] The equiaxed crystal ratio distribution was obtained based on the equiaxed crystal ratio of the ingot sample along the width direction of the ingot.

[0040] Optionally, the equiaxed crystal sample image is preprocessed to obtain a preprocessed image, including:

[0041] The equiaxed crystal sample image is subjected to grayscale transformation, tilt correction, image denoising and image scaling to obtain a preprocessed image.

[0042] Optionally, the system further comprises a data interface and a continuous casting secondary module,

[0043] The equiaxed crystal image acquisition module is used to obtain an equiaxed crystal sample image of the equiaxed crystal sample;

[0044] The data interface is used to connect the equiaxed crystal image acquisition module, the continuous casting secondary module and the equiaxed crystal intelligent detection and grading module to achieve data communication between the equiaxed crystal image acquisition module, the continuous casting secondary module and the equiaxed crystal intelligent detection and grading module;

[0045] The continuous casting secondary module is used to record data in the continuous casting production process.

[0046] In a third aspect, an embodiment of the present invention further provides a computer storage medium on which a computer program is stored, and when the program is executed by a processor, the steps of any of the above methods are implemented.

[0047] The beneficial effects of the embodiments of this specification are as follows:

[0048] The embodiments of this specification provide a solution, by obtaining an equiaxed crystal sample image of an equiaxed crystal sample, segmenting the equiaxed crystal sample image of the equiaxed crystal sample through a pre-trained semantic segmentation model, obtaining an equiaxed crystal area, measuring the equiaxed crystal area, obtaining the equiaxed crystal rate and the equiaxed crystal rate distribution; judging the equiaxed crystal performance of the equiaxed crystal sample based on the equiaxed crystal rate and the equiaxed crystal rate distribution. By using image processing technology and semantic segmentation technology, intelligent and accurate automatic segmentation of the equiaxed crystal area of ​​the equiaxed crystal sample is achieved, and combined with computer graphics measurement technology, automatic measurement of the equiaxed crystal area, automatic calculation of the equiaxed crystal rate and evaluation of the equiaxed crystal rate distribution are achieved. Further, based on the equiaxed crystal rate and the equiaxed crystal rate distribution, the equiaxed crystal performance of the equiaxed crystal sample is judged, and intelligent judgment of the equiaxed crystal performance of the sample is achieved, solving the problems of poor efficiency, low accuracy, poor identity and unscientificity in the existing manual measurement of equiaxed crystals. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present specification. Also, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0050] Figure 1A flow chart of a method for intelligent identification and performance determination of equiaxed crystals in a casting billet provided in an embodiment of this specification;

[0051] Figure 2 A schematic diagram of a system for intelligently identifying and determining the performance of equiaxed crystals in a casting billet provided in an embodiment of this specification.

[0052] Figure 3 A flowchart of semantic segmentation of an equiaxed crystal region provided in an embodiment of this specification.

[0053] Figure 4 This is a working principle diagram of an equiaxed crystal grading module provided in an embodiment of this specification.

[0054] Figure 5 It is a schematic diagram of a block structure of an electronic device provided by an embodiment of the present invention.

[0055] Markings in the figure: bus 500; receiver 501; processor 502; transmitter 503; memory 504; bus interface 505; equiaxed crystal image acquisition module 1; data interface 2; continuous casting secondary module 3; equiaxed crystal intelligent detection and grading module 4; equiaxed crystal semantic segmentation module 401; image preprocessing module 4011; equiaxed crystal segmentation data set 4012; segmentation network model 4013; equiaxed crystal measurement module 402; equiaxed crystal rate calculation module 4201; equiaxed crystal distribution module 4202; ingot parameter module 403; steel type 4031; width 4032; pulling speed 4033; thickness 4034; ingot number 4035; equiaxed crystal grading module 404; equiaxed crystal grading rules 4041; equiaxed crystal grading model 4042; data storage module 4043. DETAILED DESCRIPTION

[0056] In order to better understand the above technical scheme, the technical scheme of the embodiments of this specification is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of this specification and the specific features in the embodiments are detailed descriptions of the technical scheme of the embodiments of this specification, rather than limitations on the technical scheme of this specification. In the absence of conflict, the embodiments of this specification and the technical features in the embodiments can be combined with each other.

[0057] like Figure 1 As shown, it is a flow chart of a method for intelligent identification and performance determination of equiaxed crystals of a casting billet provided in an embodiment of this specification, and the method comprises the following steps:

[0058] S101: Acquire an equiaxed crystal sample image of an equiaxed crystal sample.

[0059] S102: Segmenting the equiaxed crystal sample image of the equiaxed crystal sample using a pre-trained semantic segmentation model to obtain an equiaxed crystal region.

[0060] S103: measuring the equiaxed crystal region to obtain the equiaxed crystal ratio and the equiaxed crystal ratio distribution.

[0061] In the embodiment of the present invention, the equiaxed crystal region may be measured using computer graphics measurement technology.

[0062] S104: Determine the equiaxed crystal performance of the equiaxed crystal sample based on the equiaxed crystal ratio and the equiaxed crystal ratio distribution.

[0063] Through image processing technology and semantic segmentation technology, intelligent and accurate automatic segmentation of the equiaxed crystal area in the equiaxed crystal sample is achieved. Combined with computer graphics measurement technology, automatic measurement of the equiaxed crystal area, automatic calculation of the equiaxed crystal ratio and evaluation of the equiaxed crystal ratio distribution are achieved. The equiaxed crystal performance of the equiaxed crystal sample is further determined based on the equiaxed crystal ratio and equiaxed crystal ratio distribution, thereby achieving intelligent judgment of the equiaxed crystal performance of the sample, and solving the problems of poor efficiency, low accuracy, poor identity and unscientificity in the existing manual measurement of equiaxed crystals.

[0064] As a further step, the training method of the semantic segmentation model is:

[0065] The collected equiaxed crystal sample image is preprocessed to obtain a preprocessed image.

[0066] An equiaxed crystal sample image training set is obtained, wherein the equiaxed crystal sample image training set includes a plurality of unlabeled preprocessed images and a plurality of labeled equiaxed crystal region preprocessed images.

[0067] The equiaxed crystal segmentation network model is trained based on the equiaxed crystal sample image training set.

[0068] When the loss function of the equiaxed crystal segmentation network model converges or the number of training times reaches a predetermined value, the training of the equiaxed crystal segmentation network model is terminated to obtain a trained equiaxed crystal segmentation network model.

[0069] In an embodiment of the present invention, the equiaxed crystal segmentation network model may be a semantic segmentation model, such as a fully convolutional neural network (FCN).

[0070] As a further step, the equiaxed crystal region is measured to obtain the equiaxed crystal ratio and equiaxed crystal ratio distribution, including:

[0071] The equiaxed crystal area of ​​the sample and the total area of ​​the sample are obtained; the equiaxed crystal area of ​​the sample is the area of ​​the equiaxed crystal area in the equiaxed crystal sample image, and the total area of ​​the sample is the area occupied by the sample ingot in the equiaxed crystal sample image.

[0072] Based on the equiaxed crystal region area and the total sample area, the equiaxed crystal ratio is obtained.

[0073] Among them, equiaxed crystal ratio = equiaxed crystal area / total sample area.

[0074] The equiaxed crystal ratio distribution was obtained based on the equiaxed crystal ratio of the ingot sample along the width direction of the ingot.

[0075] In the embodiment of the present invention, while segmenting the equiaxed crystal sample image of the equiaxed crystal sample using a pre-trained semantic segmentation model, the coordinate information of the equiaxed crystal region contour is extracted.

[0076] Specifically, the equiaxed crystal ratio distribution is obtained based on the equiaxed crystal ratio of the ingot sample along the width direction of the ingot:

[0077] The horizontal axis area x∈[x1,x2], the values ​​of x1 and x2 can be arbitrary numbers, and can also be limited to real numbers, x1 is less than or equal to x2, the area of ​​the equiaxed crystal region is s1, and the equiaxed crystal rate of this region is s1 / ((x2-x1)*h), where h is the thickness of the sample. The distribution is discrete along the width of the sample. For example, the equiaxed crystal rate of the interval is calculated every 5cm on the horizontal axis, and the width direction is divided into several 5cm areas. After calculating the corresponding equiaxed crystal rate, all discrete points and the corresponding equiaxed crystal rate are fitted, and the distribution of the equiaxed crystal rate along the width of the sample can be obtained.

[0078] As a further step, the equiaxed crystal sample image is preprocessed to obtain a preprocessed image, including:

[0079] The equiaxed crystal sample image is subjected to grayscale transformation, tilt correction, image denoising and image scaling to obtain a preprocessed image.

[0080] As a further step, the equiaxed crystal performance of the equiaxed crystal sample is determined based on the equiaxed crystal rate and the equiaxed crystal rate distribution, which can be: if the equiaxed crystal rate is greater than the first set value, the grade of the ingot sample is determined to be a superior product. If the equiaxed crystal rate is greater than the second set value and less than the first set value, the grade of the ingot sample is determined to be a medium product. If the equiaxed crystal rate is less than the second set value, the grade of the ingot sample is determined to be a second-rate product. The value of the second set value is smaller than the first set value, and the value of the second set value can be 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, and the value of the first set value can be 0.5, 0.6, 0.7, 0.8, 0.9.

[0081] Optionally, the equiaxed crystal performance of the equiaxed crystal sample is determined based on the equiaxed crystal rate and the equiaxed crystal rate distribution. Alternatively, based on the equiaxed crystal detection data and the experience of on-site detection technicians, equiaxed crystal grading rules are formulated, and an equiaxed crystal grading model is established to intelligently determine the equiaxed crystal performance of the sample. The equiaxed crystal detection data includes equiaxed crystal rate, equiaxed crystal rate distribution, overall equiaxed crystal rate, equiaxed crystal distribution, maximum equiaxed crystal rate, minimum equiaxed crystal rate, secondary equiaxed crystal and other data.

[0082] In the embodiment of the present invention, the data involved in the above method will also be stored.

[0083] Example 2

[0084] Based on the above-mentioned method for intelligently identifying and determining the performance of equiaxed crystals in a casting billet, an embodiment of the present invention provides an execution subject for executing the above-mentioned method for intelligently identifying and determining the performance of equiaxed crystals in a casting billet, and the execution subject is a system for intelligently identifying and determining the performance of equiaxed crystals in a casting billet. The system includes an equiaxed crystal image acquisition module and an equiaxed crystal intelligent detection and grading module.

[0085] The equiaxed crystal image acquisition module is used to obtain the equiaxed crystal sample image of the equiaxed crystal sample.

[0086] The equiaxed crystal intelligent detection and grading module includes an equiaxed crystal semantic segmentation module, an equiaxed crystal measurement module, a billet parameter module, and an equiaxed crystal grading module.

[0087] The equiaxed crystal semantic segmentation module is used to segment the equiaxed crystal sample image of the equiaxed crystal sample through a pre-trained semantic segmentation model to obtain the equiaxed crystal area.

[0088] The equiaxed crystal measurement module is used to measure the equiaxed crystal area to obtain the equiaxed crystal ratio and equiaxed crystal ratio distribution.

[0089] The casting parameter module is used to store the process parameters of casting production.

[0090] The equiaxed crystal grading module is used to judge and store data on the equiaxed crystal performance of the equiaxed crystal sample based on the equiaxed crystal ratio and the equiaxed crystal ratio distribution.

[0091] Optionally, the system further includes a data interface and a continuous casting secondary module.

[0092] The equiaxed crystal image acquisition module is used to obtain the equiaxed crystal sample image of the equiaxed crystal sample.

[0093] The data interface is used to connect the equiaxed crystal image acquisition module, the continuous casting secondary module and the equiaxed crystal intelligent detection and grading module to realize data communication among the equiaxed crystal image acquisition module, the continuous casting secondary module and the equiaxed crystal intelligent detection and grading module.

[0094] The continuous casting secondary module is used to record data during the continuous casting production process.

[0095] Optionally, the equiaxed crystal semantic segmentation module is further used for:

[0096] The collected equiaxed crystal sample image is preprocessed to obtain a preprocessed image.

[0097] An equiaxed crystal sample image training set is obtained, wherein the equiaxed crystal sample image training set includes a plurality of unlabeled preprocessed images and a plurality of labeled equiaxed crystal region preprocessed images.

[0098] The equiaxed crystal segmentation network model is trained based on the equiaxed crystal sample image training set.

[0099] When the loss function of the equiaxed crystal segmentation network model converges or the number of training times reaches a predetermined value, the training of the equiaxed crystal segmentation network model is terminated to obtain a trained equiaxed crystal segmentation network model.

[0100] As a further step, the equiaxed crystal measurement module is specifically used for:

[0101] The equiaxed crystal area of ​​the sample and the total area of ​​the sample are obtained; the equiaxed crystal area of ​​the sample is the area of ​​the equiaxed crystal area in the equiaxed crystal sample image, and the total area of ​​the sample is the area occupied by the sample ingot in the equiaxed crystal sample image.

[0102] Based on the equiaxed crystal region area and the total sample area, the equiaxed crystal ratio is obtained.

[0103] The equiaxed crystal ratio distribution was obtained based on the equiaxed crystal ratio of the ingot sample along the width direction of the ingot.

[0104] As an optional embodiment, the following is combined Figure 2 The intelligent identification and performance determination system of equiaxed crystals in ingots is explained.

[0105] The ingot equiaxed crystal intelligent identification and performance judgment system consists of four parts: equiaxed crystal image acquisition module 1, data interface 2, continuous casting secondary module 3 and equiaxed crystal intelligent detection and grade judgment module 4.

[0106] The equiaxed crystal image acquisition module 1 is used to obtain the end face image of the equiaxed crystal sample;

[0107] The data interface 2 is used to realize data communication among the equiaxed crystal image acquisition device 1 , the continuous casting secondary module 3 and the equiaxed crystal intelligent detection and grading module 4 .

[0108] The data interface 2 may be a data communication mode such as TCP / IP, PROFIBUS, or RS485.

[0109] The above-mentioned continuous casting secondary module 3 is used to record data in the continuous casting production process, such as continuous casting process data, planning and scheduling, production performance data, equipment sensor data, etc.

[0110] The above-mentioned equiaxed crystal intelligent detection and grading module 4 includes an equiaxed crystal semantic segmentation module 401, an equiaxed crystal measurement module 402, a casting blank parameter module 403, and an equiaxed crystal grading module 404. The equiaxed crystal semantic segmentation module 401 segments the equiaxed crystal region in the equiaxed crystal sample through the segmentation network model 4013; the equiaxed crystal measurement module 402 measures the segmented equiaxed crystal region through the equiaxed crystal rate calculation module 4021, automatically calculates the equiaxed crystal rate, and uses the equiaxed crystal distribution module 4202 to quantitatively evaluate the equiaxed crystal rate distribution; the casting blank parameter module 403 is used to store the process parameters of the casting blank production; the equiaxed crystal grading module 404 performs intelligent judgment on the equiaxed crystal performance of the sample through the equiaxed crystal grading model 4042, and the data storage module 4043 completes the automatic data storage.

[0111] The equiaxed crystal semantic segmentation module 401 pre-processes the collected equiaxed crystal sample images through the image pre-processing module 4011; completes the labeling of the equiaxed crystal area in the sample image, and produces the equiaxed crystal segmentation data set 4012.

[0112] The casting parameter module 403 is used to store the required process parameters such as steel type 4031, width 4032, casting speed 4033, thickness 4034, and billet number 4035.

[0113] The equiaxed crystal grading module 404 formulates equiaxed crystal grading rules 4041 based on the equiaxed crystal detection data and the experience of on-site detection technicians.

[0114] The working principle of the intelligent identification and performance determination system for equiaxed crystals of ingots provided by the embodiment of the present invention is as follows:

[0115] 1. The equiaxed crystal image acquisition module 1 is used to obtain the end face image of the equiaxed crystal sample and transmit the sample image to the equiaxed crystal semantic segmentation module 401 through the data interface 2.

[0116] 2. The equiaxed crystal semantic segmentation module 401 performs grayscale transformation, tilt correction, image denoising, image scaling and other processing on the equiaxed crystal sample image through the image preprocessing module 4011. The equiaxed crystal region in the preprocessed sample image is manually annotated to produce an equiaxed crystal segmentation data set 4012. An equiaxed crystal segmentation network model 4013 is constructed, and the equiaxed crystal segmentation network model 4013 is trained through the equiaxed crystal segmentation data set 4012 to complete the structure and parameter optimization of the equiaxed crystal segmentation network model 4013 and obtain an ideal equiaxed crystal segmentation network model 4013. The trained equiaxed crystal segmentation model 4013 is used to perform equiaxed crystal region segmentation on the equiaxed crystal sample image to be detected to obtain the pixel coordinates of the equiaxed crystal region.

[0117] 3. The equiaxed crystal measurement module 402 is based on the equiaxed crystal area contour of the semantic segmentation. By using computer graphics measurement technology, through the equiaxed crystal calculation module 4021, the equiaxed crystal area and the total area of ​​the sample can be easily obtained, and the equiaxed crystal ratio can be automatically calculated. The equiaxed crystal distribution module 4022 is used to obtain the equiaxed crystal ratio of the ingot sample along the width direction of the ingot, and then the uniformity of the equiaxed crystal is evaluated.

[0118] 4. The ingot parameter module 403 collects data such as the steel type 4031, width 4032, casting speed 4033, thickness 4034, ingot number 4035, etc. of the continuous casting secondary module 3 through the data interface 2, and completes data storage and real-time update.

[0119] 5. Based on the equiaxed crystal detection data and the experience of on-site detection technicians, the equiaxed crystal grading module 404 formulates equiaxed crystal grading rules 4041 and establishes an equiaxed crystal grading model 4042, performs intelligent judgment on the equiaxed crystal performance of the sample, stores the data generated in the process through the data storage module 4043, and transmits the equiaxed crystal core detection and grading data to the continuous casting secondary module 3.

[0120] As a further reference, Figure 3 An equiaxed crystal semantic segmentation module 401 provided in an embodiment of the present invention is described as follows:

[0121] The equiaxed crystal semantic segmentation module 401 obtains the equiaxed crystal sample image through the data interface 2, and uses the image preprocessing 4011 to perform grayscale transformation, tilt correction, image denoising, image scaling and other processing on the equiaxed crystal sample image; the equiaxed crystal area in the preprocessed sample image is manually labeled to produce an equiaxed crystal segmentation data set 4012. The equiaxed crystal segmentation network model 4013 is constructed, and the equiaxed crystal segmentation network model 4013 is trained through the equiaxed crystal segmentation data set 4012 to complete the structure and parameter optimization of the equiaxed crystal segmentation network model 4013, and obtain the ideal equiaxed crystal segmentation network model 4013. The trained equiaxed crystal segmentation model 4013 is used to perform equiaxed crystal area segmentation on the equiaxed crystal sample image to be detected, and the pixel coordinates of the equiaxed crystal area are obtained.

[0122] Reference Figure 4 An equiaxed crystal grading module 404 provided in an embodiment of the present invention is described as follows:

[0123] Based on the historical data of equiaxed crystal detection and the experience of on-site detection technicians, variables such as overall equiaxed crystal rate, equiaxed crystal distribution, maximum equiaxed crystal rate, minimum equiaxed crystal rate, and secondary equiaxed crystal are used as the basis for judgment. Combined with the continuous casting production process parameters, an equiaxed crystal grading rule 4041 is formulated, and the grading rule optimization is completed through rule verification; an equiaxed crystal grading model 4042 is established according to the equiaxed crystal grading rule 4041, and the equiaxed crystal performance test results of the sample are input into the equiaxed crystal grading model 4042 to complete the intelligent judgment of the equiaxed crystal performance of the sample, and the data generated in the process are stored through the data storage module 4043, and the equiaxed crystal core detection and grading data are transmitted to the continuous casting secondary module 3.

[0124] Regarding the above-mentioned device, the specific functions of each module therein have been described in detail in the embodiment of the method for intelligent identification and performance determination of equiaxed crystals of cast billets provided in the embodiments of this specification, and will not be elaborated here.

[0125] Based on the same inventive concept as the method for intelligently identifying equiaxed crystals of a casting and determining performance in the aforementioned embodiment, the embodiment of this specification further provides an electronic device, such as Figure 5 As shown, it includes a memory 504, a processor 502, and a computer program stored in the memory 504 and executable on the processor 502. When the processor 502 executes the program, the steps of any of the methods for intelligent identification and performance determination of equiaxed crystals of ingots described above are implemented.

[0126] Among them, Figure 5 In the embodiment of the present invention, a bus architecture (represented by bus 500) is shown, which may include any number of interconnected buses and bridges, and bus 500 links various circuits including one or more processors represented by processor 502 and memory represented by memory 504. Bus 500 may also link various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. Bus interface 505 provides an interface between bus 500 and receiver 501 and transmitter 503. Receiver 501 and transmitter 503 may be the same element, namely a transceiver, providing a unit for communicating with various other devices over a transmission medium. Processor 502 is responsible for managing bus 500 and general processing, while memory 504 may be used to store data used by processor 502 when performing operations.

[0127] Based on the inventive concept of the method for intelligent identification and performance determination of equiaxed crystals in cast billets in the aforementioned embodiments, the embodiments of this specification also provide a computer-readable storage medium on which a computer program is stored, which, when executed by a processor, implements the steps of any of the methods for intelligent identification and performance determination of equiaxed crystals in cast billets described above.

[0128] This specification is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that has the functions specified in one or more boxes.

[0129] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0130] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0131] Although the preferred embodiments of this specification have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of this specification.

[0132] Obviously, those skilled in the art can make various changes and modifications to this specification without departing from the spirit and scope of this specification. Thus, if these modifications and variations of this specification fall within the scope of the claims of this specification and their equivalents, this specification is also intended to include these modifications and variations.

Claims

1. A method for intelligent identification and performance determination of equiaxed crystals of ingots, characterized in that: The method comprises: obtaining an equiaxed crystal sample image of the equiaxed crystal sample; The equiaxed crystal sample image of the equiaxed crystal sample is segmented by a pre-trained semantic segmentation model to obtain the equiaxed crystal region; The equiaxed crystal region is measured to obtain the equiaxed crystal ratio and equiaxed crystal ratio distribution; The equiaxed crystal properties of the equiaxed crystal sample are determined based on the equiaxed crystal ratio and the equiaxed crystal ratio distribution; The training method of the semantic segmentation model is: Preprocessing the collected equiaxed crystal sample image to obtain a preprocessed image; Obtaining an equiaxed crystal sample image training set, wherein the equiaxed crystal sample image training set includes a plurality of unlabeled preprocessed images and a plurality of labeled equiaxed crystal region preprocessed images; Training an equiaxed crystal segmentation network model based on the equiaxed crystal sample image training set; When the loss function of the equiaxed crystal segmentation network model converges or the number of training times reaches a predetermined value, the training of the equiaxed crystal segmentation network model is terminated to obtain a trained equiaxed crystal segmentation network model; The measuring of the equiaxed crystal region to obtain the equiaxed crystal ratio and the equiaxed crystal ratio distribution includes: The equiaxed crystal area of ​​the sample and the total area of ​​the sample are obtained; the equiaxed crystal area of ​​the sample is the area of ​​the equiaxed crystal area in the equiaxed crystal sample image, and the total area of ​​the sample is the area occupied by the sample ingot in the equiaxed crystal sample image; Based on the area of ​​the equiaxed crystal region and the total area of ​​the sample, obtaining an equiaxed crystal ratio; The equiaxed crystal ratio distribution is obtained based on the equiaxed crystal ratio of the ingot sample along the width direction of the ingot; The method of obtaining the equiaxed crystal ratio distribution based on the equiaxed crystal ratio of the ingot sample along the width direction of the ingot includes: dividing the width direction into a plurality of regions, respectively obtaining the corresponding equiaxed crystal ratios, fitting all discrete points and the corresponding equiaxed crystal ratios, and obtaining the distribution of the equiaxed crystal ratio along the width of the sample.

2. The method according to claim 1, characterized in that The equiaxed crystal sample image is preprocessed to obtain a preprocessed image, including: The equiaxed crystal sample image is subjected to grayscale transformation, tilt correction, image denoising and image scaling to obtain a preprocessed image.

3. A system for intelligent identification and performance determination of equiaxed crystals of ingots, the system being used to implement the steps of the method according to any one of claims 1 to 2, characterized in that: The system includes an equiaxed crystal image acquisition module and an equiaxed crystal intelligent detection and grading module; wherein, The equiaxed crystal image acquisition module is used to obtain an equiaxed crystal sample image of the equiaxed crystal sample; The equiaxed crystal intelligent detection and grading module includes an equiaxed crystal semantic segmentation module, an equiaxed crystal measurement module, a casting blank parameter module, and an equiaxed crystal grading module; The equiaxed crystal semantic segmentation module is used to segment the equiaxed crystal sample image of the equiaxed crystal sample by using a pre-trained semantic segmentation model to obtain an equiaxed crystal region; The equiaxed crystal measurement module is used to measure the equiaxed crystal area to obtain the equiaxed crystal ratio and equiaxed crystal ratio distribution; The casting parameter module is used to store the process parameters of casting production; The equiaxed crystal grading module is used to determine the equiaxed crystal performance of the equiaxed crystal sample based on the equiaxed crystal ratio and the equiaxed crystal ratio distribution.

4. The system according to claim 3, characterized in that The isometric crystal semantic segmentation module is also used for: Preprocessing the collected equiaxed crystal sample image to obtain a preprocessed image; Obtaining an equiaxed crystal sample image training set, wherein the equiaxed crystal sample image training set includes a plurality of unlabeled preprocessed images and a plurality of labeled equiaxed crystal region preprocessed images; Training an equiaxed crystal segmentation network model based on the equiaxed crystal sample image training set; When the loss function of the equiaxed crystal segmentation network model converges or the number of training times reaches a predetermined value, the training of the equiaxed crystal segmentation network model is terminated to obtain a trained equiaxed crystal segmentation network model.

5. The system according to claim 3, characterized in that The equiaxed crystal measurement module is specifically used for: The equiaxed crystal area of ​​the sample and the total area of ​​the sample are obtained; the equiaxed crystal area of ​​the sample is the area of ​​the equiaxed crystal area in the equiaxed crystal sample image, and the total area of ​​the sample is the area occupied by the sample ingot in the equiaxed crystal sample image; Based on the area of ​​the equiaxed crystal region and the total area of ​​the sample, obtaining an equiaxed crystal ratio; The equiaxed crystal ratio distribution was obtained based on the equiaxed crystal ratio of the ingot sample along the width direction of the ingot.

6. The system according to claim 3, characterized in that The equiaxed crystal sample image is preprocessed to obtain a preprocessed image, including: The equiaxed crystal sample image is subjected to grayscale transformation, tilt correction, image denoising and image scaling to obtain a preprocessed image.

7. The system according to claim 3, characterized in that The system also includes a data interface and a continuous casting secondary module, The equiaxed crystal image acquisition module is used to obtain an equiaxed crystal sample image of the equiaxed crystal sample; The data interface is used to connect the equiaxed crystal image acquisition module, the continuous casting secondary module and the equiaxed crystal intelligent detection and grading module to achieve data communication between the equiaxed crystal image acquisition module, the continuous casting secondary module and the equiaxed crystal intelligent detection and grading module; The continuous casting secondary module is used to record data in the continuous casting production process.

8. A computer storage medium, characterized in that: A computer program is stored thereon, and when the program is executed by a processor, the steps of the method according to any one of claims 1 to 2 are implemented.

Images