阅读说明：本技术 可自动调节放肩工艺参数的单晶炉及控制方法 (Single crystal furnace capable of automatically adjusting shouldering process parameters and control method ) 是由 逯占文 邢治国 徐永胜 王学卫 王胜军 张靖 于 2020-08-25 设计创作，主要内容包括：本发明提供一种可自动调节放肩工艺参数的单晶炉,包括：晶体提拉模块、机器视觉直径测量模块和温度控制模块；所述提拉系统包括：转轴、连接在转轴上的绕线轮和一端连接在绕线轮上的软轴；所述软轴的另一端与晶体连接,在拉制单晶的工作过程中,转轴带动绕线轮旋转,使所述软轴与绕线轮相连的一端缠绕在绕线轮上,进而使软轴与晶体相连的一端对晶体进行提拉；所述提拉系统通过驱动电机穿过减速器,进而通过有齿轮齿条机构带动。本发明能使晶体放肩生长过程在整个生长过程中始终保持放肩形状一致,在不同单晶炉内部生长环境下,有效调整放肩生长的工艺参数。(The invention provides a single crystal furnace capable of automatically adjusting shouldering process parameters, which comprises: the device comprises a crystal pulling module, a machine vision diameter measuring module and a temperature control module; the lifting system comprises: the device comprises a rotating shaft, a reel connected to the rotating shaft and a flexible shaft with one end connected to the reel; the other end of the flexible shaft is connected with the crystal, and in the working process of pulling the single crystal, the rotating shaft drives the reel to rotate, so that one end, connected with the reel, of the flexible shaft is wound on the reel, and the other end, connected with the crystal, of the flexible shaft is used for pulling the crystal; the lifting system passes through the speed reducer through the driving motor and is driven by the gear rack mechanism. The invention can keep the shouldering shape of the crystal in the whole growth process and effectively adjust the technological parameters of the shouldering growth under different growth environments in the single crystal furnace.)

1. A single crystal furnace capable of automatically adjusting shouldering process parameters is characterized by comprising: the device comprises a crystal pulling module, a machine vision diameter measuring module and a temperature control module;

the lifting system comprises: the device comprises a rotating shaft, a reel connected to the rotating shaft and a flexible shaft with one end connected to the reel; the other end of the flexible shaft is connected with the crystal, and in the working process of pulling the single crystal, the rotating shaft drives the reel to rotate, so that one end, connected with the reel, of the flexible shaft is wound on the reel, and the other end, connected with the crystal, of the flexible shaft is used for pulling the crystal; the lifting system passes through the speed reducer through the driving motor and is driven by the gear rack mechanism; the lifting system realizes lifting motion through a motor and simultaneously drives a rolling spline shaft through a turbine speed reducer, and the rolling spline shaft pair drives a winding wheel to rotate and bear support;

the machine vision diameter measuring module captures a single crystal growth image through an industrial camera, and the image acquired by the industrial camera is converted into an image signal and a digital signal and is transmitted to an upper computer in real time to acquire the edge position of the single crystal.

2. A control method of a single crystal furnace capable of automatically adjusting shouldering process parameters, which applies the single crystal furnace capable of automatically adjusting shouldering process parameters according to claim 1, is characterized by comprising the following steps:

s1: in the crystal shoulder-setting growth process, obtaining SP values of a group of growth control parameters from an SOP standard operation guidance parameter table;

s2: inputting a PID automatic regulation generation parameter group, correcting an OP value and obtaining the OP value; the OP value is set in a control module for the shouldering growth of the crystal to control the shouldering growth of the crystal;

s3: acquiring a diameter value measured by a diameter measuring module, a power value measured by a temperature control module and a pulling speed value generated by a warning pulling system;

s4: generating an information base for automatically adjusting shouldering process parameters through the classification abstraction and the analysis processing unified processing of data, and further generating a knowledge base through a neural network algorithm;

s5: therefore, a group of technological parameters capable of controlling shouldering growth are automatically generated according to actually generated growth data through the rules and characteristics of the knowledge base.

3. The method for controlling a single crystal furnace capable of automatically adjusting shouldering process parameters according to claim 2, wherein,

the data acquisition and control module acquires real-time original growth data of the single crystal furnace, and the original growth data form an n-dimensional array; firstly, carrying out standardization processing on the n-dimensional data set; constructing a covariance matrix:

wherein n represents the total number of data, uj and uk represent the mean values of features j and k, respectively; since the mean of the samples is 0 in the normalized data, the covariance of the normalized data can be expressed as:

forming a matrix W with (d, k) dimension by the selected k eigenvectors; in the k characteristic extraction processes, selecting variables as characteristic values to complete the standardization and dimension reduction processing of the n-dimensional array;

the matrix W is a data set of the shoulder-laying growth control information, namely an information base of data is constructed;

by adopting a 3-layer feedback neural network, wherein the first layer is an input unit, the second layer is called a hidden layer, the third layer is called an output layer, the data of the information base is classified, and a control algorithm realizes a training sample set by the following five steps:

taking a sample { Ai, Bi } from the set of samples, where Ai is the output and Bi is the desired output; calculating the actual output, oii, of the network; calculating the error D ═ B_i-О_i；

Adjusting the connection weight matrix W of each hidden layer according to the difference of the actual output error D; the connection weight matrix W adjusts the formula of the hidden layers to be:

W_ij(t+1)＝W_ij(t)+a(B_i-О_i)x_j(t)；

wherein, W_ijRepresenting the connection weights between neurons i to j; a represents a constant of learning speed; x_jRepresenting a current state of the neuron; and the feature vector of the input unit of the first layer is k, and the number of nodes of the second layer called hidden layer is set to be k +1 or k + 3.

Technical Field

The invention relates to the technical field of automatic adjustment shouldering processes, in particular to a single crystal furnace capable of automatically adjusting shouldering process parameters and a control method.

Background

At present, a single crystal furnace is the most widely applied crystal growth equipment, and a Czochralski method (CZ) is also the most widely used crystal growth method with the longest research history and the most mature process. However, many factors still influence the crystal yield, quality and yield of the crystal in the crystal growth process, so that the significance of researching the crystal growth control and each sub-growth process of the crystal growth and improving the crystal survival rate is very important.

When the crystal is grown by the Czochralski method, the process parameters for controlling the crystal growth are an important part of the crystal growth. The crystal quality and yield of the crystal can be greatly different according to different parameter settings of various single crystal furnaces. The crystal shouldering growth process mentioned in the invention is a key process for crystal growth and crystallization. The success or failure of the shoulder growth of the crystal directly influences the success or failure of the whole crystal growth. The control method (ACCP) is an improved activity of crystal shouldering algorithm of a control system in a single crystal furnace control system. The control algorithm is used for calculating, and the automatically generated control parameters interact with the control system to control the shouldering growth process, so that the acquisition method and the control technology of the shouldering process parameters are improved. The ACCP method provides a good system control strategy for the control system. The control method effectively utilizes the crystal growth control parameters such as crystal growth speed, shouldering initial speed, shouldering diameter starting measurement length, power regulation coefficient and the like, records the crystal growth data in real time according to the crystal growth, and constructs a knowledge base of the shouldering control parameters on the basis of the data, thereby overcoming the defects of the shouldering method of the traditional control system and the popular angle shouldering method at present. The traditional shouldering algorithms of the existing single crystal furnace control system are different in size, and the common problems are that the control process is complex, the algorithm control is difficult, the control process parameter compiling difficulty is high, the crystal growth internal environment is complex and various, manual intervention is needed in the growth process, a whole set of mature process parameters is difficult to exist, and the existing complex and variable crystal internal growth environment can be adapted.

The traditional crystal shouldering control algorithm generally adjusts shouldering parameters according to the temperature of crystal growth and the pull rate of actual crystal growth, and can be simply understood as a shouldering algorithm with a fixed pull rate. At present, although a plurality of people propose an angle shouldering control algorithm, the growth control problem of the traditional shouldering process can be improved to a certain extent, the problems of too much manual intervention, high parameter compiling difficulty, poor adaptability to the internal environment of actual crystal growth and the like are also common and outstanding, and the problems are influenced by professional qualities of process technicians, so that a great amount of manual participation is required to intervene in the shouldering growth process. If the intervention is excessive or not timely, the shape of the shoulder of the crystal is poor, the survival rate is high, and the labor cost is high. In the actual crystal growth process of the traditional angle shouldering algorithm, a plurality of uncertain factors often exist, and the operation difficulty of a single crystal furnace and the risk of crystal breakage are increased.

Disclosure of Invention

According to the technical problems of operation difficulty of the single crystal furnace and high risk of crystal breakage, the single crystal furnace capable of automatically adjusting shouldering process parameters and the control method are provided. The invention mainly utilizes a single crystal furnace capable of automatically adjusting shouldering process parameters, which is characterized by comprising the following steps: the device comprises a crystal pulling module, a machine vision diameter measuring module and a temperature control module.

The lifting system comprises: the device comprises a rotating shaft, a reel connected to the rotating shaft and a flexible shaft with one end connected to the reel; the other end of the flexible shaft is connected with the crystal, and in the working process of pulling the single crystal, the rotating shaft drives the reel to rotate, so that one end, connected with the reel, of the flexible shaft is wound on the reel, and the other end, connected with the crystal, of the flexible shaft is used for pulling the crystal; the lifting system passes through the speed reducer through the driving motor and is driven by the gear rack mechanism; the lifting system realizes lifting motion through the motor and simultaneously drives the rolling spline shaft through the turbine speed reducer, and the rolling spline shaft pair drives the rolling wheel to rotate and take up support.

Further, the machine vision diameter measuring module captures the image of the single crystal growth through an industrial camera, and the image acquired by the industrial camera is converted into an image signal and a digital signal, and is transmitted to an upper computer in real time to acquire the edge position of the single crystal.

The invention also comprises a control method of the single crystal furnace capable of automatically adjusting shouldering process parameters, which is characterized by comprising the following steps:

step S1: in the crystal shoulder-setting growth process, obtaining SP values of a group of growth control parameters from an SOP standard operation guidance parameter table;

step S2: inputting a PID automatic regulation generation parameter group, correcting an OP value and obtaining the OP value; the OP value is set in a control module for the shouldering growth of the crystal to control the shouldering growth of the crystal;

step S3: acquiring a diameter value measured by a diameter measuring module, a power value measured by a temperature control module and a pulling speed value generated by a warning pulling system;

step S4: generating an information base for automatically adjusting shouldering process parameters through the classification abstraction and the analysis processing unified processing of data, and further generating a knowledge base through a neural network algorithm;

step S5: therefore, a group of technological parameters capable of controlling shouldering growth are automatically generated according to actually generated growth data through the rules and characteristics of the knowledge base.

Furthermore, the data acquisition and control module acquires real-time original growth data of the single crystal furnace, and the original growth data form an n-dimensional array; firstly, carrying out standardization processing on the n-dimensional data set; constructing a covariance matrix:

wherein n represents the total number of data, uj and uk represent the mean values of features j and k, respectively; since the mean of the samples is 0 in the normalized data, the covariance of the normalized data can be expressed as:

forming a matrix W with (d, k) dimension by the selected k eigenvectors; in the k characteristic extraction processes, selecting variables as characteristic values to complete the standardization and dimension reduction processing of the n-dimensional array;

the matrix W is a data set of the shoulder-laying growth control information, namely an information base of data is constructed;

by adopting a 3-layer feedback neural network, wherein the first layer is an input unit, the second layer is called a hidden layer, the third layer is called an output layer, the data of the information base is classified, and a control algorithm realizes a training sample set by the following five steps:

taking a sample { Ai, Bi } from the set of samples, where Ai is the output and Bi is the desired output; calculating the actual output, oii, of the network; calculating the error D ═ B_i-О_i；

Adjusting the connection weight matrix W of each hidden layer according to the difference of the actual output error D; the connection weight matrix W adjusts the formula of the hidden layers to be:

W_ij(t+1)＝W_ij(t)+a(B_i-О_i)x_j(t)；

wherein, W_ijRepresenting the connection weights between neurons i to j; a represents a constant of learning speed; x_jRepresenting a current state of the neuron; and the feature vector of the input unit of the first layer is k, and the number of nodes of the second layer called hidden layer is set to be k +1 or k + 3.

Compared with the prior art, the invention has the following advantages:

the invention designs a single crystal furnace with automatic adjustment of shouldering process parameters and a control method based on the control method of the original pulling single crystal furnace starting from the source of the process parameter control of the crystal shouldering growth process. The shouldering shape of the crystal can be kept consistent all the time in the whole growth process of the crystal shouldering growth process, and the technological parameters of the shouldering growth can be effectively adjusted under different growth environments in the single crystal furnace.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.

FIG. 1 is a schematic diagram of a control device according to the present invention.

FIG. 2 is a schematic diagram of the control of the shoulder growth of the crystal according to the present invention.

FIG. 3 is a schematic diagram of a subsystem for automatically adjusting shouldering process parameters according to the present invention.

Detailed Description

In order to make the technical solutions of the present invention better understood, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

It should be noted that the terms "first," "second," and the like in the description and claims of the present invention and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the invention described herein are capable of operation in sequences other than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

As shown in fig. 1 to 3, the single crystal furnace capable of automatically adjusting shouldering process parameters according to the present invention is characterized by comprising: the device comprises a crystal pulling module, a machine vision diameter measuring module and a temperature control module. The lifting system passes through the speed reducer through the driving motor and is driven by the gear rack mechanism; the lifting system realizes lifting motion through a motor and simultaneously drives a rolling spline shaft through a turbine speed reducer, and the rolling spline shaft pair drives a winding wheel to rotate and bear support; the machine vision diameter measuring module captures a single crystal growth image through an industrial camera, and an image system acquired by the industrial camera converts an image signal and a digital signal, transmits the image signal and the digital signal to an upper computer in real time and acquires the edge position of the single crystal.

In the present application, as a preferred embodiment, the diameter of the drawing system can be controlled by changing the speed of the drawing. As the diameter becomes larger, a suitable increase in the pull rate may reduce the diameter. Conversely, as the diameter becomes smaller, decreasing the pull rate may increase the diameter. In the control process, when the temperature control system enables the temperature in the furnace body to be overhigh, the crystallization speed of the crystal is slow, and the crystallization speed is fast when the temperature is overlow. Therefore, the pulling system, the machine vision diameter measurement module and the temperature control system, which are interactive with respect to crystal growth. That is, crystal growth requires effective adjustment of their control parameters to meet the desired crystal growth requirements.

Further, the invention also comprises a control method of the single crystal furnace capable of automatically adjusting shouldering process parameters, which is characterized by comprising the following steps:

step S1: in the crystal shoulder-setting growth process, obtaining SP values of a group of growth control parameters from an SOP standard operation guidance parameter table;

step S2: inputting a PID automatic regulation generation parameter group, correcting an OP value and obtaining the OP value; the OP value is set in a control module for the shouldering growth of the crystal to control the shouldering growth of the crystal;

step S3: an automatic adjusting mechanism is introduced in the process of controlling the shoulder of the crystal, the diameter value measured by a diameter measuring module is obtained, the power value measured by a temperature control module and the pulling speed value generated by a vigilant pulling system are collected;

step S4: generating an information base for automatically adjusting shouldering process parameters through the classification abstraction and the analysis processing unified processing of data, and further generating a knowledge base through a neural network algorithm;

step S5: therefore, a group of technological parameters capable of controlling shouldering growth are automatically generated according to actually generated growth data through the rules and characteristics of the knowledge base.

The diameter machine vision measuring system measures the numerical value of the change of the shouldering diameter in the process of shouldering the crystal in real time, and inputs the numerical value to the control of the shouldering angle so as to obtain the adjusting parameter required to be controlled. And adjusting a temperature control module of the system according to the angle value so as to control the crystal growth temperature in the single crystal furnace. The temperature will affect the speed of the lifting or lowering of the pulling equipment of the single crystal furnace. Through the interaction of the control modules, the internal growth condition of the single crystal furnace during crystal growth is improved. The control method of the control system of the single crystal furnace can better automatically adjust the growth parameters under the condition that the crystal growth environments are different, so that the shoulder of the grown crystal is consistent with the process parameter table in the shouldering growth process of the crystal, the survival rate of the crystal is further ensured, and the productivity of the single crystal furnace is improved.

The subsystem for automatically adjusting the shouldering process parameters is realized as follows. The data acquisition and control module can acquire a large amount of real-time original growth data of the single crystal furnace every day. The data is an n-dimensional array formed by the acquisition module according to the data collection mode of the system. This n-dimensional dataset is first normalized. Constructing a covariance matrix:

wherein n represents the total number of data, uj and uk represent the mean values of features j and k, respectively; in the normalized data, the mean of the samples is 0, so the covariance of the normalized data can be represented again as:

and forming a matrix W with (d, k) dimensions by the selected k eigenvectors. In the k feature extraction process, variables such as temperature, pulling speed, power and diameter are selected as feature values. This completes the normalization and dimension reduction processing of the n-dimensional array. The W matrix is a data set which is generated by people and has certain information capable of controlling shoulder growth, namely an information base for constructing and completing data. In order to enhance the efficiency of storing the data information into the database, a module for accelerating the storage of the information base of the data is added in the process of storing the data information into the database.

The construction of knowledge bases is the process of filtering, refining and processing the relevant information to obtain useful data. Knowledge itself may also yield new knowledge through autonomous and non-autonomous learning. The invention adopts a 3-layer feedback neural network, wherein the first layer is an input unit, the second layer is called a hidden layer, and the third layer is called an output layer. And (4) carrying out classification processing on the data of the information base, and controlling an algorithm to realize a training sample set in the following five steps.

Taking a sample { Ai, Bi } from the set of samples, where Ai is the output and Bi is the desired output;

calculating the actual output, oii, of the network;

calculating the error D ═ B_i-О_i；

Adjusting the connection weight matrix W of each hidden layer according to the difference of the actual output error D; the above process is repeated for each sample until the error sigma does not exceed a controllable range for the entire sample set. In the neural network designed by the invention, the first-layer input unit is a feature vector k selected by people. The number of nodes in the second layer, referred to as the hidden layer, is set to k +1 or k + 3. Experimental test and measurement results show that the k value has little influence on the recognition rate, but the excessive number of nodes increases the calculation amount, so that the training is slow. The parameters of the third layer, called as the output layer, are the process control parameters required by the control method of the invention, namely the growth speed of the growing crystal, the initial casting speed, the initial diameter measurement length, the power regulation coefficient and the like are controlled, and the parameters are cooperated to act on the casting stage of the crystal to achieve better control effect. In order to avoid the influence on the identification accuracy caused by the supersaturation of the neural network due to the passing of the samples, the number of the selected input sample sets is 50, wherein 45 data samples are used as a training set, and 5 data samples are used for verification.

The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments.

In the above embodiments of the present invention, the descriptions of the respective embodiments have respective emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to related descriptions of other embodiments.

Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present invention.