CN117187943B - Melt detection method and device in crystal pulling process and electronic equipment - Google Patents
Melt detection method and device in crystal pulling process and electronic equipment Download PDFInfo
- Publication number
- CN117187943B CN117187943B CN202311167997.5A CN202311167997A CN117187943B CN 117187943 B CN117187943 B CN 117187943B CN 202311167997 A CN202311167997 A CN 202311167997A CN 117187943 B CN117187943 B CN 117187943B
- Authority
- CN
- China
- Prior art keywords
- image
- molten material
- melting
- solid
- labeling
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 84
- 239000013078 crystal Substances 0.000 title claims abstract description 26
- 238000001514 detection method Methods 0.000 title claims abstract description 15
- 239000012768 molten material Substances 0.000 claims abstract description 111
- 238000002844 melting Methods 0.000 claims abstract description 81
- 239000007788 liquid Substances 0.000 claims abstract description 75
- 230000008018 melting Effects 0.000 claims abstract description 75
- 230000000877 morphologic effect Effects 0.000 claims abstract description 66
- 238000003709 image segmentation Methods 0.000 claims abstract description 29
- 239000007787 solid Substances 0.000 claims abstract description 25
- 239000000155 melt Substances 0.000 claims abstract description 17
- 238000010309 melting process Methods 0.000 claims abstract description 14
- 238000012545 processing Methods 0.000 claims abstract description 10
- 238000002372 labelling Methods 0.000 claims description 67
- 238000012549 training Methods 0.000 claims description 45
- 238000010438 heat treatment Methods 0.000 claims description 14
- 238000005266 casting Methods 0.000 claims description 13
- 238000004364 calculation method Methods 0.000 claims description 9
- 238000007781 pre-processing Methods 0.000 claims description 6
- 238000003860 storage Methods 0.000 claims description 4
- 238000004590 computer program Methods 0.000 claims description 3
- 238000004519 manufacturing process Methods 0.000 abstract description 5
- 230000007547 defect Effects 0.000 abstract description 2
- 239000002210 silicon-based material Substances 0.000 description 17
- 239000010408 film Substances 0.000 description 15
- 239000000463 material Substances 0.000 description 5
- 230000011218 segmentation Effects 0.000 description 5
- 239000010409 thin film Substances 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 2
- XUIMIQQOPSSXEZ-UHFFFAOYSA-N Silicon Chemical compound [Si] XUIMIQQOPSSXEZ-UHFFFAOYSA-N 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 238000004801 process automation Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 229910052710 silicon Inorganic materials 0.000 description 1
- 239000010703 silicon Substances 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
Landscapes
- Crystals, And After-Treatments Of Crystals (AREA)
Abstract
The application discloses a melt detection method, device and electronic equipment in crystal pulling process, including: obtaining an image to be detected in a crucible during a melting process; processing the image to be detected based on the image segmentation model to obtain melt morphological characteristics corresponding to the image to be detected and image pixel information corresponding to each melt morphological characteristic; calculating based on image pixel information corresponding to each melting stock morphological feature to obtain a solid-liquid ratio parameter; and determining the target stage of the molten material according to the solid-liquid ratio parameter. The method and the device have the advantages that the molten material morphological characteristics at least comprise one or more of solid state characteristics, film state characteristics or liquid state characteristics, the molten material morphological characteristics can be automatically identified based on the image segmentation model, the defect of manual identification is overcome, the accuracy of follow-up control is improved, and further the production efficiency is improved.
Description
Technical Field
The application relates to the technical field of photovoltaic and semiconductor single crystal pulling, in particular to a melt detection method and device in a crystal pulling process and electronic equipment.
Background
In the melting process in the field of single crystal growth, repeated casting is required to be carried out for multiple times at high temperature and vacuum state according to the melting condition of silicon materials in a single crystal furnace. In the melting process, a technician usually controls the time point of reducing the power of the bottom heater by virtue of the experience of the molten state of the silicon material, because subjective factors of the technician easily cause different occasions of reducing the power of the bottom heater, and a large amount of technicians are required to pay attention to and operate or miss the occasions of reducing the power of the bottom heater at any time, so that the production cost, the yield and the production efficiency are affected to a certain extent.
Disclosure of Invention
In view of this, the present application provides the following technical solutions:
a method for detecting molten material during crystal pulling, comprising:
obtaining an image to be detected in a crucible during a melting process;
processing the image to be detected based on an image segmentation model to obtain a molten material morphological feature corresponding to the image to be detected and image pixel information corresponding to each molten material morphological feature, wherein the image segmentation model is a model which is based on training of an overall process image from the beginning of molten material to the end of full molten material, can detect the molten material morphological feature of the image and the image pixel information corresponding to each molten material morphological feature, and the molten material morphological feature at least comprises one or more of solid state features, film state features or liquid state features;
calculating based on image pixel information corresponding to each melting stock morphological feature to obtain a solid-liquid ratio parameter;
and determining a target stage where the molten material is located according to the solid-liquid ratio parameter.
Optionally, the method further comprises:
and determining control information corresponding to the target stage of the molten material, wherein the control information is at least used for controlling the re-casting time or the heating power in the molten material process.
Optionally, the method further comprises:
obtaining an image data set, wherein the image data set comprises a plurality of images obtained by acquiring a whole process image from the beginning of melting to the end of full melting by an industrial camera end;
performing feature labeling on each image in the image data set to obtain labeling information, wherein the labeling information comprises one or more of first labeling information representing solid molten materials, second labeling information representing film molten materials and third labeling information representing liquid molten materials;
generating an image training sample based on the labeling information of each image in the image dataset and the image dataset;
and training the image training sample by taking the image characteristics of each image in the image dataset as training characteristics and the labeling information of each image as a training target to obtain an image segmentation model.
Optionally, the feature labeling of each image in the image dataset to obtain labeling information includes:
performing image enhancement preprocessing on each image in the image dataset to obtain a preprocessed image dataset;
and carrying out feature labeling on each image in the preprocessed image data set to obtain labeling information.
Optionally, the calculating based on the image pixel information corresponding to each melting stock morphological feature to obtain the solid-liquid ratio parameter includes:
obtaining a weight parameter corresponding to each melt morphological feature and the number of pixels in an image area corresponding to each melt morphological feature at the current moment;
and calculating to obtain a solid-liquid ratio parameter according to the weight parameter corresponding to each melting stock morphological feature and the corresponding pixel number.
Optionally, the determining the target stage of the molten material according to the solid-liquid ratio parameter includes:
obtaining a threshold parameter of each stage in which the molten material is located;
and determining a target stage where the molten material is located based on the threshold parameter and the solution duty ratio parameter.
A melt detection apparatus in a crystal pulling process, comprising:
the image acquisition unit is used for acquiring an image to be detected in the crucible in the melting process;
the image segmentation model is a model which is based on training the whole process image from the beginning of melting to the end of melting and can detect the melting morphological characteristics of the image and the image pixel information corresponding to each melting morphological characteristic, and the melting morphological characteristics at least comprise one or more of solid state characteristics, film state characteristics or liquid state characteristics;
the parameter calculation unit is used for calculating based on the image pixel information corresponding to the morphological characteristics of each melting stock to obtain a solid-liquid ratio parameter;
and the stage determining unit is used for determining the target stage where the molten material is located according to the solid-liquid ratio parameter.
Optionally, the method further comprises:
and the control information determining unit is used for determining control information corresponding to the target stage where the molten material is located, and the control information is at least used for controlling the re-casting time or the heating power in the molten material process.
A readable storage medium having a computer program stored thereon, which when executed by a processor, implements a method of melt detection in a crystal pulling process as defined in any one of the preceding claims.
An electronic device, comprising:
a memory for storing an application program and data generated by the operation of the application program;
a processor for executing the application program to implement the melt detection method in a crystal pulling process as set forth in any one of the preceding claims.
According to the technical scheme, the application discloses a method and a device for detecting melting materials in a crystal pulling process and electronic equipment, wherein the method comprises the following steps: obtaining an image to be detected in a crucible during a melting process; processing the image to be detected based on the image segmentation model to obtain melt morphological characteristics corresponding to the image to be detected and image pixel information corresponding to each melt morphological characteristic; calculating based on image pixel information corresponding to each melting stock morphological feature to obtain a solid-liquid ratio parameter; and determining the target stage of the molten material according to the solid-liquid ratio parameter. The image segmentation model is a model which is based on training an overall process image from the beginning of melting to the end of full melting, and can detect the melting morphological characteristics of the image and the image pixel information corresponding to each melting morphological characteristic, and the melting morphological characteristics at least comprise one or more of solid state characteristics, film state characteristics or liquid state characteristics.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings that are needed in the description of the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only embodiments of the present application, and that other drawings can be obtained according to the provided drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic flow chart of a method for detecting molten material during crystal pulling according to an embodiment of the present disclosure;
FIG. 2 is a schematic view of images before and after image segmentation of a three-modality frit region according to an embodiment of the present application;
FIG. 3 is a schematic diagram of a device for detecting molten material during a crystal pulling process according to an embodiment of the present application.
Detailed Description
The following description of the embodiments of the present application will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all, of the embodiments of the present application. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the present disclosure, are within the scope of the present disclosure.
The terms first and second and the like in the description and in the claims of the present application and in the above-described figures are used for distinguishing between different objects and not for describing a particular sequential order. Furthermore, the terms "comprise" and "have," as well as any variations thereof, are intended to cover a non-exclusive inclusion. For example, a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to the listed steps or elements but may include steps or elements not expressly listed.
In order to facilitate the description of the control method of the fusion of the seed crystal in the crystal pulling process provided in the embodiments of the present application, the related terms will now be explained.
Pulling: refers to the previous working procedure in the photovoltaic field, and the silicon material is produced into a silicon rod in a single crystal furnace by a Czochralski method.
And (3) full melting: refers to melting silicon material to be liquid by using a crucible peripheral heater.
And (3) re-casting: in the melting stage, the silicon material is insufficient, but because the solid silicon material occupies the crucible in a large volume, the silicon material needs to be melted to a specified stage to be supplemented.
Heating power: when the melting material is indicated, the working power of the heater is provided at the two sides and the bottom of the crucible.
The embodiment of the application provides a molten material detection method in a crystal pulling process, which is used for solving the problem of insufficient automation in a molten material judging process stage and realizing the whole process automation control of a molten material in a molten material stage, heating power adjustment and re-throwing time point control from the beginning of molten material to the full melting of silicon material.
Referring to fig. 1, a flow chart of a method for detecting molten material in a crystal pulling process according to an embodiment of the present application may include the following steps:
s101, obtaining an image to be detected in the crucible in the melting process.
And acquiring an overall process image from the beginning of melting to the end of full melting through an industrial camera end, and taking the image in the crucible acquired at the current moment during the melting process as an image to be detected.
S102, processing the image to be detected based on an image segmentation model to obtain melt morphological characteristics corresponding to the image to be detected and image pixel information corresponding to each melt morphological characteristic.
The image segmentation model is a model which is obtained based on training of an overall process image from the beginning of melting to the end of full melting, and can detect melting morphological characteristics of the image and image pixel information corresponding to each melting morphological characteristic, wherein the melting morphological characteristics at least comprise one or more of solid state characteristics, film state characteristics or liquid state characteristics.
The embodiment of the application also provides a method for generating the image segmentation model, which comprises the following steps:
obtaining an image data set, wherein the image data set comprises a plurality of images obtained by acquiring a whole process image from the beginning of melting to the end of full melting by an industrial camera end; performing feature labeling on each image in the image data set to obtain labeling information, wherein the labeling information comprises one or more of first labeling information representing solid molten materials, second labeling information representing film molten materials and third labeling information representing liquid molten materials; generating an image training sample based on the labeling information of each image in the image dataset and the image dataset; and training the image training sample by taking the image characteristics of each image in the image dataset as training characteristics and the labeling information of each image as a training target to obtain an image segmentation model.
Specifically, in the process of generating the image training sample, the whole process image from the beginning of melting to the end of the whole melting can be acquired through the industrial camera end (for example, the whole process image n sheets of multiple melting forms are acquired at intervals of m seconds). In order to make the image features more obvious, image enhancement preprocessing may be performed on each image in the image dataset, resulting in a preprocessed image dataset. Then, the images in the preprocessed image dataset are subjected to state division, and different types of annotation extraction work is performed on the molten material characteristics in each state.
Specifically, when the CCD camera captures an image in the crucible in the melting process, the image is transmitted to an image feature segmentation model for image processing, and the melting form in the image is divided into three types: solid state, film state and liquid state, and calculating the number of the pixel points occupied by each solid state, film state and liquid state. The invention mainly performs operations such as scaling, translation, rotation and the like on the original image to realize the expansion of the number of pictures on the basis of unchanged characteristic content of the original image, and avoids poor model learning effect caused by insufficient number of images.
The labeling operation of the image is to label the characteristics of the preprocessed image, and labeling information is as follows: solid-state melt (for example, can be first annotation information, specifically can be annotated to 0), film-state melt (for example, can be second annotation information, specifically can be annotated to 1), liquid melt (for example, can be third annotation information, specifically can be annotated to 2), the annotation frame requires to laminate the melt edge, because the melt form is often various, need all annotate out with the silicon material form that exists in the training sample when annotating, is equivalent to increasing sample quantity to training model result is more outstanding. Compared with the prior art, the method for dividing the molten material morphology is finer, compared with the method for dividing the solid-liquid molten material according to the pixel threshold value and the connection area, the method is not easily affected by production environment, the solid-liquid ratio is more favorable for the next weighted calculation of the solid-liquid ratio because the image display area is only often displayed on the surface of the liquid surface, the solid-state silicon material is often immersed in the liquid molten material in a large part of volume, the thin-film silicon material is often small in volume, and if the solid-liquid molten material and the solid-liquid molten material are divided according to the pixel threshold value and the connection area, the judgment of the molten material stage is often inaccurate.
And S103, calculating based on image pixel information corresponding to each melting stock morphological feature to obtain a solid-liquid ratio parameter.
S104, determining a target stage where the molten material is located according to the solid-liquid ratio parameter.
After determining the target stage in the embodiment of the present application, a control instruction may be generated according to the stage where the current molten material is located, or a corresponding parameter threshold may be set according to the target stage, so as to automatically control a subsequent process flow. Further, the information corresponding to the target stage can be fed back to the control system, the number of the pixels is weighted, and whether feeding re-feeding and bottom heating are needed to be performed or not is judged according to the ratio result. Specifically, the embodiment of the application further includes: and determining control information corresponding to the target stage of the molten material, wherein the control information is at least used for controlling the re-casting time or the heating power in the molten material process. The target stage may include a stage from the beginning of melting to the non-melting stage of the semi-melting setting, a stage of semi-melting but not full-melting, and a stage of full-melting, which may be further divided according to actual requirements.
In one implementation manner of the embodiment of the present application, the calculating based on the image pixel information corresponding to each molten material morphological feature to obtain the solid-liquid ratio parameter includes: obtaining a weight parameter corresponding to each melt morphological feature and the number of pixels in an image area corresponding to each melt morphological feature at the current moment; and calculating to obtain a solid-liquid ratio parameter according to the weight parameter corresponding to each melting stock morphological feature and the corresponding pixel number.
Specifically, the determining, according to the solid-liquid ratio parameter, the target stage where the molten material is located includes: obtaining a threshold parameter of each stage in which the molten material is located; and determining a target stage where the molten material is located based on the threshold parameter and the solution duty ratio parameter.
In the embodiment of the application, the solid-liquid ratio is determined through weighted calculation, wherein the weighted calculation solid-liquid ratio is to calculate the number of pixel points in a molten material area with three forms of solid, film and liquid respectively, and judge the molten material stage according to the calculation result and whether to give a control command to close bottom heating and re-casting material. Fig. 2 shows images before and after image division of three forms of molten material regions, and in the lower image division output image, a molten material solid state in a blue region, a molten material thin film state in a yellow region, and a molten material liquid state in a green region. And (3) calculating the number of the pixels, and after receiving the image processed by the image segmentation model, the upper computer (or the processing module) calculates the number of the pixels of each color area according to the divided fixed area color pixel values. For example: A. b, C are weights of solid state, film state and liquid state respectively, and if the melting stock at a certain moment detects that the number of pixels occupied by the solid state is a, the number of pixels occupied by the film state is b and the number of pixels occupied by the liquid state is c. And then setting A, B, C coefficients according to different process stages of a weighted relation formula with solid-liquid ratio d of d=aa+Bb+cc/a+b+c. The process stage is often divided according to whether or not the half melting or the full melting is achieved, for example, the molten material is started to the stage of not reaching the half melting set unmelted material 0, the stage of reaching the half melting but not reaching the full melting set molten material 1, and the conditions for judging the half melting (the moment of closing the bottom heating at this time) are set to be that the solid state ratio is 50%, the full melting ratio is 100% and the liquid state ratio is 30%. Because there is often much large volume of silicon material in the frit 0 stage and much small volume of silicon material is immersed below, the frit 0 stage solid silicon material coefficient is set to 1.5, the film state is 1, and the liquid state is 0.5. When the half melting is achieved, there is usually no larger volume of silicon material, but still part of silicon material is below the liquid level, and the full melting is to calculate whether the liquid state ratio is 100%, so the coefficients are set to be solid 1.2, film 0.7 and liquid 1. The coefficient and the judging semi-molten and full-molten solid-liquid ratio threshold can be adjusted according to actual conditions, so that the flexibility is good, and the accuracy is high.
According to the embodiment of the application, through introducing an image segmentation technology applied to the image field by deep learning, the invention discloses a melt detection method based on image target feature segmentation, and based on target feature segmentation information, a weighted formula is utilized to calculate the solid-liquid ratio, so that the defect caused by manual operation can be effectively avoided, the time for turning off or reducing a bottom heater and performing re-casting is controlled more accurately, and the half-melting time and the full-melting time are judged. Further, in the application, the characteristic state model is output through collecting batch melting process images and performing preprocessing operation, then performing characteristic labeling and extracting operation on the preprocessed images, and then training; in actual production, image segmentation models are called in real time to carry out reasoning analysis of target characteristics on image data of a camera end, so that areas of melting materials in three forms and pixel point information are divided, the pixel point information is subjected to weighted calculation to obtain solid-liquid ratio, and accordingly whether a solid-liquid ratio value reaches a preset value or not is judged according to the solid-liquid ratio value, and further feeding re-casting can be effectively controlled, bottom heating is closed, and half melting and full melting moments are judged.
In another implementation manner of the embodiment of the present application, there is also provided a melt detection apparatus in a crystal pulling process, referring to fig. 3, including:
an image obtaining unit 201 for obtaining an image to be detected in the crucible at the time of the melting process;
the model processing unit 202 is configured to process the image to be detected based on an image segmentation model to obtain a molten material morphological feature corresponding to the image to be detected and image pixel information corresponding to each molten material morphological feature, where the image segmentation model is a model capable of detecting the molten material morphological feature of the image and the image pixel information corresponding to each molten material morphological feature, and the molten material morphological feature at least includes one or more of a solid state feature, a thin film state feature, or a liquid state feature, based on training of a whole-process image from the beginning of molten material to the end of whole molten material;
a parameter calculation unit 203, configured to calculate based on image pixel information corresponding to each melt morphology feature, to obtain a solid-liquid ratio parameter;
and the stage determining unit 204 is configured to determine a target stage in which the molten material is located according to the solid-liquid ratio parameter.
Optionally, the method further comprises:
and the control information determining unit is used for determining control information corresponding to the target stage where the molten material is located, and the control information is at least used for controlling the re-casting time or the heating power in the molten material process.
Optionally, the apparatus further comprises: a model training unit, the model training unit comprising:
the data acquisition subunit is used for acquiring an image data set, wherein the image data set comprises a plurality of images acquired by the industrial camera end for acquiring a whole process image from the beginning of melting to the end of full melting;
the characteristic labeling subunit is used for carrying out characteristic labeling on each image in the image data set to obtain labeling information, wherein the labeling information comprises one or more of first labeling information for representing solid molten materials, second labeling information for representing thin-film molten materials and third labeling information for representing liquid molten materials;
a sample generation subunit, configured to generate an image training sample based on the labeling information of each image in the image dataset and the image dataset;
and the training subunit is used for taking the image characteristics of each image in the image dataset as training characteristics, taking the labeling information of each image as a training target, and training the image training sample to obtain an image segmentation model.
Optionally, the feature labeling subunit is specifically configured to:
performing image enhancement preprocessing on each image in the image dataset to obtain a preprocessed image dataset;
and carrying out feature labeling on each image in the preprocessed image data set to obtain labeling information.
Optionally, the parameter calculation unit is specifically configured to:
obtaining a weight parameter corresponding to each melt morphological feature and the number of pixels in an image area corresponding to each melt morphological feature at the current moment;
and calculating to obtain a solid-liquid ratio parameter according to the weight parameter corresponding to each melting stock morphological feature and the corresponding pixel number.
Optionally, the stage determining unit includes:
obtaining a threshold parameter of each stage in which the molten material is located;
and determining a target stage where the molten material is located based on the threshold parameter and the solution duty ratio parameter.
According to the melt detection device in the crystal pulling process, through learning the whole process image of the silicon material melt stage, three melt states are divided, the melt target features of each state are marked, then the features are extracted and trained to obtain the target feature segmentation model, the melt forms existing in the current melt image are analyzed in real time through the target feature segmentation model in a reasoning mode, the respective areas of the melt forms are segmented, the stage where the melt is located is obtained, then the stage where the melt is located is further fed back to a control system, the number of pixels of the stage is calculated in a weighting mode, whether the feeding re-casting and the bottom heating closing are carried out is judged according to the ratio result, and the processing accuracy and efficiency are improved.
It should be noted that, the specific implementation of each unit and sub-unit in this embodiment may refer to the corresponding content in the foregoing, which is not described in detail herein.
In another embodiment of the present application, there is also provided a readable storage medium having stored thereon a computer program which, when executed by a processor, implements a method of melt detection in a crystal pulling process as set forth in any one of the preceding claims.
In another embodiment of the present application, there is also provided an electronic device, which may include:
a memory for storing an application program and data generated by the operation of the application program;
a processor for executing the application program to realize:
obtaining an image to be detected in a crucible during a melting process;
processing the image to be detected based on an image segmentation model to obtain a molten material morphological feature corresponding to the image to be detected and image pixel information corresponding to each molten material morphological feature, wherein the image segmentation model is a model which is based on training of an overall process image from the beginning of molten material to the end of full molten material, can detect the molten material morphological feature of the image and the image pixel information corresponding to each molten material morphological feature, and the molten material morphological feature at least comprises one or more of solid state features, film state features or liquid state features;
calculating based on image pixel information corresponding to each melting stock morphological feature to obtain a solid-liquid ratio parameter;
and determining a target stage where the molten material is located according to the solid-liquid ratio parameter.
Optionally, the method further comprises:
and determining control information corresponding to the target stage of the molten material, wherein the control information is at least used for controlling the re-casting time or the heating power in the molten material process.
Optionally, the method further comprises:
obtaining an image data set, wherein the image data set comprises a plurality of images obtained by acquiring a whole process image from the beginning of melting to the end of full melting by an industrial camera end;
performing feature labeling on each image in the image data set to obtain labeling information, wherein the labeling information comprises one or more of first labeling information representing solid molten materials, second labeling information representing film molten materials and third labeling information representing liquid molten materials;
generating an image training sample based on the labeling information of each image in the image dataset and the image dataset;
and training the image training sample by taking the image characteristics of each image in the image dataset as training characteristics and the labeling information of each image as a training target to obtain an image segmentation model.
Optionally, the feature labeling of each image in the image dataset to obtain labeling information includes:
performing image enhancement preprocessing on each image in the image dataset to obtain a preprocessed image dataset;
and carrying out feature labeling on each image in the preprocessed image data set to obtain labeling information.
Optionally, the calculating based on the image pixel information corresponding to each melting stock morphological feature to obtain the solid-liquid ratio parameter includes:
obtaining a weight parameter corresponding to each melt morphological feature and the number of pixels in an image area corresponding to each melt morphological feature at the current moment;
and calculating to obtain a solid-liquid ratio parameter according to the weight parameter corresponding to each melting stock morphological feature and the corresponding pixel number.
Optionally, the determining the target stage of the molten material according to the solid-liquid ratio parameter includes:
obtaining a threshold parameter of each stage in which the molten material is located;
and determining a target stage where the molten material is located based on the threshold parameter and the solution duty ratio parameter.
It should be noted that, the specific implementation of the processor in this embodiment may refer to the corresponding content in the foregoing, which is not described in detail herein.
In the present specification, each embodiment is described in a progressive manner, and each embodiment is mainly described in a different point from other embodiments, and identical and similar parts between the embodiments are all enough to refer to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant points refer to the description of the method section.
Those of skill would further appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative elements and steps are described above generally in terms of functionality in order to clearly illustrate the interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.
The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. The software modules may be disposed in Random Access Memory (RAM), memory, read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (9)
1. A method for detecting molten material during a crystal pulling process, comprising:
obtaining an image to be detected in a crucible during a melting process;
processing the image to be detected based on an image segmentation model to obtain a molten material morphological feature corresponding to the image to be detected and image pixel information corresponding to each molten material morphological feature, wherein the image segmentation model is a model which is based on training of an overall process image from the beginning of molten material to the end of full molten material, can detect the molten material morphological feature of the image and the image pixel information corresponding to each molten material morphological feature, and the molten material morphological feature at least comprises one or more of solid state features, film state features or liquid state features; the image segmentation model generation process is configured to: obtaining an image data set, wherein the image data set comprises a plurality of images obtained by acquiring a whole process image from the beginning of melting to the end of full melting by an industrial camera end; performing feature labeling on each image in the image data set to obtain labeling information, wherein the labeling information comprises one or more of first labeling information representing solid molten materials, second labeling information representing film molten materials and third labeling information representing liquid molten materials; generating an image training sample based on the labeling information of each image in the image dataset and the image dataset; taking the image characteristics of each image in the image dataset as training characteristics, taking the labeling information of each image as a training target, and training the image training sample to obtain an image segmentation model;
calculating based on image pixel information corresponding to each melting stock morphological feature to obtain a solid-liquid ratio parameter;
and determining a target stage where the molten material is located according to the solid-liquid ratio parameter.
2. The method as recited in claim 1, further comprising:
and determining control information corresponding to the target stage of the molten material, wherein the control information is at least used for controlling the re-casting time or the heating power in the molten material process.
3. The method of claim 1, wherein the feature labeling each image in the image dataset to obtain labeling information comprises:
performing image enhancement preprocessing on each image in the image dataset to obtain a preprocessed image dataset;
and carrying out feature labeling on each image in the preprocessed image data set to obtain labeling information.
4. The method of claim 1, wherein the calculating based on the image pixel information corresponding to each of the melt morphology features to obtain the solid-liquid ratio parameter comprises:
obtaining a weight parameter corresponding to each melt morphological feature and the number of pixels in an image area corresponding to each melt morphological feature at the current moment;
and calculating to obtain a solid-liquid ratio parameter according to the weight parameter corresponding to each melting stock morphological feature and the corresponding pixel number.
5. The method of claim 1, wherein determining the target stage at which the frit is located based on the solid-liquid ratio parameter comprises:
obtaining a threshold parameter of each stage in which the molten material is located;
and determining a target stage where the molten material is located based on the threshold parameter and the solid-liquid ratio parameter.
6. A melt detection device in a crystal pulling process, comprising:
the image acquisition unit is used for acquiring an image to be detected in the crucible in the melting process;
the image segmentation model is a model which is based on training the whole process image from the beginning of melting to the end of melting and can detect the melting morphological characteristics of the image and the image pixel information corresponding to each melting morphological characteristic, and the melting morphological characteristics at least comprise one or more of solid state characteristics, film state characteristics or liquid state characteristics; the image segmentation model generation process is configured to: obtaining an image data set, wherein the image data set comprises a plurality of images obtained by acquiring a whole process image from the beginning of melting to the end of full melting by an industrial camera end; performing feature labeling on each image in the image data set to obtain labeling information, wherein the labeling information comprises one or more of first labeling information representing solid molten materials, second labeling information representing film molten materials and third labeling information representing liquid molten materials; generating an image training sample based on the labeling information of each image in the image dataset and the image dataset; taking the image characteristics of each image in the image dataset as training characteristics, taking the labeling information of each image as a training target, and training the image training sample to obtain an image segmentation model;
the parameter calculation unit is used for calculating based on the image pixel information corresponding to the morphological characteristics of each melting stock to obtain a solid-liquid ratio parameter;
and the stage determining unit is used for determining the target stage where the molten material is located according to the solid-liquid ratio parameter.
7. The apparatus as recited in claim 6, further comprising:
and the control information determining unit is used for determining control information corresponding to the target stage where the molten material is located, and the control information is at least used for controlling the re-casting time or the heating power in the molten material process.
8. A readable storage medium having stored thereon a computer program which, when executed by a processor, implements the method of melt detection in a crystal pulling process as defined in any one of claims 1-5.
9. An electronic device, comprising:
a memory for storing an application program and data generated by the operation of the application program;
a processor for executing the application program to implement the melt detection method in a crystal pulling process as defined in any one of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311167997.5A CN117187943B (en) | 2023-09-11 | 2023-09-11 | Melt detection method and device in crystal pulling process and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311167997.5A CN117187943B (en) | 2023-09-11 | 2023-09-11 | Melt detection method and device in crystal pulling process and electronic equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117187943A CN117187943A (en) | 2023-12-08 |
| CN117187943B true CN117187943B (en) | 2024-02-27 |
Family
ID=89001186
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311167997.5A Active CN117187943B (en) | 2023-09-11 | 2023-09-11 | Melt detection method and device in crystal pulling process and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117187943B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117468083B (en) * | 2023-12-27 | 2024-05-28 | 浙江晶盛机电股份有限公司 | Control method and device for seed crystal lowering process, crystal growth furnace system and computer equipment |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112488980A (en) * | 2019-08-20 | 2021-03-12 | 隆基绿能科技股份有限公司 | Molten material state detection method, device and equipment |
| CN113215651A (en) * | 2021-04-08 | 2021-08-06 | 银川隆基光伏科技有限公司 | Crystal pulling control method and equipment, single crystal furnace and computer storage medium |
| CN115992381A (en) * | 2022-12-28 | 2023-04-21 | 浙江晶盛机电股份有限公司 | Single crystal furnace re-casting method, single crystal furnace re-casting device, computer equipment and storage medium |
| CN116043323A (en) * | 2023-01-12 | 2023-05-02 | 晶澳（邢台）太阳能有限公司 | Method for detecting melting of solid silicon material |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114419435A (en) * | 2021-12-31 | 2022-04-29 | 联想(北京)有限公司 | Detection method, device, equipment and storage medium |
-
2023
- 2023-09-11 CN CN202311167997.5A patent/CN117187943B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112488980A (en) * | 2019-08-20 | 2021-03-12 | 隆基绿能科技股份有限公司 | Molten material state detection method, device and equipment |
| CN113215651A (en) * | 2021-04-08 | 2021-08-06 | 银川隆基光伏科技有限公司 | Crystal pulling control method and equipment, single crystal furnace and computer storage medium |
| CN115992381A (en) * | 2022-12-28 | 2023-04-21 | 浙江晶盛机电股份有限公司 | Single crystal furnace re-casting method, single crystal furnace re-casting device, computer equipment and storage medium |
| CN116043323A (en) * | 2023-01-12 | 2023-05-02 | 晶澳（邢台）太阳能有限公司 | Method for detecting melting of solid silicon material |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117187943A (en) | 2023-12-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN117187943B (en) | Melt detection method and device in crystal pulling process and electronic equipment | |
| WO2022213675A1 (en) | Crystal pulling control method and device, single-crystal furnace and computer storage medium | |
| US8545623B2 (en) | Method and apparatus for controlling the growth process of a monocrystalline silicon ingot | |
| CN115125609B (en) | Sapphire seeding control process and system based on image processing algorithm | |
| CN110751669A (en) | Novel CBOCP online infrared converter tapping steel flow automatic detection and tracking method and system | |
| CN111222360B (en) | Method, equipment and storage medium for detecting molten state of silicon material | |
| CN102787353B (en) | Method for measuring crystal wire of silicon rod of single crystal furnace in non-contact mode | |
| CN115880271A (en) | Identification and detection method for seed crystal single crystal line edge angle in crystal growth process | |
| JP2007110932A (en) | Automatic method for culturing and cell culture apparatus | |
| CN106435731A (en) | Automatic seeding control system of sapphire furnace and seeding control method of single crystal furnace | |
| CN112784640A (en) | Menu making method and device and cooking machine | |
| CN113411510A (en) | Camera automatic exposure algorithm based on image quality evaluation and red hot forging | |
| CN114387251B (en) | Method for monitoring a saturation point, storage medium, terminal and crystal pulling apparatus | |
| CN114387248B (en) | Silicon material melting degree monitoring method, storage medium, terminal and crystal pulling equipment | |
| CN116449789A (en) | Intelligent control method for charging barrel and related products | |
| CN113673529B (en) | Semantic segmentation model training method, silicon fusion state detection method and electronic equipment | |
| CN111753588A (en) | Power equipment online perception and focusing algorithm based on deep learning | |
| US20240006087A1 (en) | Method and electronic device for assisting with surveillance of an element of a nuclear reactor, associated computer and associated system | |
| CN117187942B (en) | Crucible position control method and device in crystal pulling process | |
| CN113673529A (en) | Semantic segmentation model training method, silicon fusion state detection method and electronic equipment | |
| CN112488980B (en) | Melt state detection method, device and equipment | |
| CN117764962A (en) | Method, system, equipment and storage medium for automatically re-throwing silicon material of single crystal furnace | |
| WO2020103826A1 (en) | Melt material state detection method, device and equipment | |
| CN118064983A (en) | Czochralski monocrystalline silicon welding method, device, equipment and storage medium | |
| CN116523883B (en) | Intelligent latex viscosity control method based on image processing |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |