A crystal furnace pressure intelligent control method and system

Abstract

The present invention relates to the field of intelligent control technology, and in particular to a method and system for intelligent control of pressure in a crystal furnace. The method comprises the steps of: obtaining a preset local range of each pressure data point in a pressure data sequence, and obtaining the number of extreme value points in the local range; calculating the noise performance degree of the pressure data points in the pressure data sequence; using the distance between the pressure data points as a weight, and correcting the weight by the noise performance degree of the pressure data points; using the corrected weight between the pressure data points in an MST clustering outlier detection algorithm to denoise the pressure data sequence, and inputting the denoised pressure data sequence into a PID algorithm to realize the control of the pressure in the crystal furnace, thereby effectively improving the accuracy of the pressure control of the crystal furnace.

Description

Technical Field

[0001] The present invention relates to the field of intelligent control technology, and in particular to a method and system for intelligently controlling pressure in a crystal furnace. Background Art

[0002] The crystal growth furnace is a process test instrument used in the field of electronics and communication technology. When in use, the instrument places the pre-treated crystal into the high-pressure furnace, and then injects nitrogen into the high-pressure cavity to keep the crystal in the quartz tube in balance with the external pressure, thereby achieving comprehensive control of pressure and temperature during the single crystal growth process and reducing the generation of crystal twins. Therefore, the pressure of the high-pressure furnace needs to be controlled in real time during use.

[0003] In the prior art, the pressure value of the device is mainly controlled by the PID algorithm. For example, the patent application document with publication number CN118672312A discloses a pressure control method and system based on a pressure transmitter. The application sets a target pressure value; adopts a PID controller and sets pressure control parameters; monitors the pressure through the control device and obtains the real-time pressure value; compares the target pressure value and the real-time pressure value and obtains the pressure error; obtains the environmental parameters around the pressure transmitter; regulates the pressure according to the pressure error and the environmental parameters and obtains feedback information; performs a second calibration according to the feedback information and optimizes and adjusts the pressure control parameters.

[0004] Although the above existing solutions can achieve the control of equipment pressure to a certain extent, in the process of implementing PID control, there may be noise interference in the pressure data input to PID, resulting in the inability to accurately adjust the pressure value of the equipment.

[0005] Based on this, how to accurately control the pressure in the crystal furnace is an urgent problem to be solved by technical personnel in this field. Summary of the invention

[0006] In order to solve the technical problem of how to accurately control the pressure in a crystal furnace, the present invention provides a method and system for intelligently controlling the pressure in a crystal furnace.

[0007] In a first aspect, the present invention provides a method for intelligently controlling pressure in a crystal furnace, which adopts the following technical solution:

[0008] A method for intelligently controlling pressure in a crystal furnace comprises the following steps:

[0009] Obtain a preset local range of each pressure data point in the pressure data sequence, and obtain the number of extreme value points in the local range; calculate the noise performance degree of the pressure data point in the pressure data sequence:

[0010] ;

[0011] In the formula, Indicates the noise performance of the i-th pressure data point, represents the variance function, represents the variance of the Euclidean distances between all adjacent extreme points in the local range of the i-th pressure data point, represents the number of extreme points in the local range of the i-th pressure data point, , Respectively represent the jth and jth pressure data points in the local range of the i-th pressure data point The pressure value of the extreme point, Represents the absolute value sign; takes the distance between the pressure data points as the weight, and corrects the weight according to the noise performance of the pressure data points; uses the corrected weight between the pressure data points in the MST clustering outlier detection algorithm to denoise the pressure data sequence, and inputs the denoised pressure data sequence into the PID algorithm to realize the control of the pressure in the crystal furnace.

[0012] The present invention takes into account that noise data may affect the control of pressure in the crystal furnace, and therefore performs denoising by acquiring the noise performance degree of pressure data points. In this process, the present invention also takes into account that noise data in the pressure data may fluctuate, and the distance weight between pressure data points obtained based on this has low accuracy. Therefore, the weights between pressure data points are corrected by acquiring the noise performance degree of pressure data points, thereby effectively avoiding the influence of noise data fluctuations on weight calculation and improving the accuracy of crystal furnace pressure control. In addition, the present invention also takes into account that normal data may fluctuate, resulting in mixing with noise data. Therefore, the present invention corrects the noise performance degree of pressure data points through the correlation between pressure data and temperature data, thereby effectively avoiding the influence of normal data fluctuations on denoising effect, thereby effectively improving the accuracy of crystal furnace pressure control.

[0013] According to a method for intelligently controlling pressure in a crystal furnace provided by the present invention, the preset local range of each pressure data point in the pressure data sequence is obtained, and the pressure value in the crystal furnace is collected by a pressure sensor, and each pressure value is used as a pressure data point, and the pressure data sequence of the crystal furnace is formed after preprocessing.

[0014] According to a method for intelligently controlling pressure in a crystal furnace provided by the present invention, the weight is corrected by the noise performance degree of the pressure data point, and also includes: obtaining a temperature data sequence of the crystal furnace, obtaining a temperature data segment corresponding to the local range in the temperature data sequence, and determining the Pearson correlation coefficient between the pressure data segment and the temperature data segment corresponding to the local range; after correcting the noise performance degree by the Pearson correlation coefficient, the possible noise degree of the pressure data point is obtained; and the product of the absolute value of the difference between the possible noise degrees of any two pressure data points and the weight is used as the correction weight between the two pressure data points.

[0015] According to a crystal furnace pressure intelligent control method provided by the present invention, the noise level of the pressure data point satisfies the relationship:

[0016] ;

[0017] In the formula, represents the possible degree of noise of the ith pressure data point, Indicates the noise performance of the i-th pressure data point, represents the pressure data segment corresponding to the local range of the i-th pressure data point, Represents the temperature data segment corresponding to the local range of the i-th pressure data point, represents the Pearson correlation coefficient between the pressure data segment and the temperature data segment corresponding to the local range of the i-th pressure data point, Represents the normalization function.

[0018] The present invention provides a method for calculating the possible degree of noise of pressure data points accurately. By acquiring temperature data that is correlated with the pressure data, the accuracy of the noise performance degree of the obtained pressure data points can be effectively improved, and the influence of normal data fluctuations on data denoising can be effectively avoided.

[0019] According to a method for intelligently controlling pressure in a crystal furnace provided by the present invention, the correction weights between the pressure data points are used in the MST clustering outlier detection algorithm to denoise the pressure data sequence, including: taking each pressure data point in the pressure data sequence as a node, and taking the correction weights between the nodes as edges to construct a weighted graph; obtaining the number of edges between the nodes of the minimum spanning tree by acquiring the minimum spanning tree in the weighted graph; dividing the minimum spanning tree into multiple spanning subtrees by cutting off M edges with the largest correction weights, where M is a preset number of cut edges; taking nodes in the spanning subtrees whose correction weights between nodes exceed the preset weights as outliers, and eliminating them to obtain a denoised pressure data sequence.

[0020] The present invention can effectively identify and remove noise points in the data by performing denoising based on the MST clustering outlier detection algorithm. In the process of constructing MST, the algorithm retains the important relationship between pressure data points, and can also retain important information of the data sequence while denoising, thereby more accurately reflecting the changing trend and fluctuation of pressure and providing more reliable input data for the PID algorithm.

[0021] According to a method for intelligently controlling the pressure in a crystal furnace provided by the present invention, the denoised pressure data sequence is input into a PID algorithm to achieve control of the pressure in the crystal furnace, including: obtaining a target pressure value required for the crystal in the crystal furnace; inputting the denoised pressure data sequence into a PID algorithm, and the PID algorithm determines the nitrogen input amount in the crystal furnace through current pressure deviation, integral accumulation and integral prediction.

[0022] The present invention processes the crystal furnace denoised pressure data sequence through the PID algorithm, can quickly respond to changes in pressure deviation, and quickly correct the deviation by adjusting the nitrogen input, thereby achieving precise control of the pressure in the crystal furnace. It also helps to shorten the time it takes for the system to reach a stable state, thereby improving production efficiency.

[0023] According to a crystal furnace pressure intelligent control method provided by the present invention, the control of the crystal furnace pressure further includes: performing abnormal monitoring on the denoised pressure data sequence and issuing an early warning of the abnormal monitoring result.

[0024] The present invention takes into account that the PID controller may be affected by external interference, equipment failure or improper parameter setting, resulting in unsatisfactory or ineffective control effect. Through abnormal monitoring, these problems can be discovered in time and corresponding measures can be taken.

[0025] In a second aspect, the present invention provides an intelligent control system for pressure in a crystal furnace, which adopts the following technical solution:

[0026] A crystal furnace internal pressure intelligent control system includes: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned crystal furnace internal pressure intelligent control method is implemented.

[0027] By adopting the above technical solution, the above-mentioned intelligent control method of pressure in a crystal furnace is generated into a computer program and stored in a memory so as to be loaded and executed by a processor, thereby making a terminal device based on the memory and the processor for easy use.

[0028] The present invention has the following technical effects:

[0029] Based on the above technical scheme, when realizing the pressure control in the crystal furnace, the present invention corrects the weights between the pressure data points by obtaining the noise performance degree of the pressure data points, thereby effectively avoiding the influence of the noise data on the weight calculation and improving the accuracy of the crystal furnace pressure control; on this basis, the present invention also obtains temperature data that is correlated with the pressure data, and corrects the noise performance degree of the pressure data points by the correlation between the pressure data and the temperature data, thereby effectively avoiding the influence of normal data fluctuations on the denoising effect, thereby effectively improving the accuracy of the crystal furnace pressure control. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] By reading the detailed description below with reference to the accompanying drawings, the above and other purposes, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-restrictive manner, and the same or corresponding reference numerals represent the same or corresponding parts.

[0031] Figure 1 A schematic flow chart of a method for intelligently controlling pressure in a crystal furnace provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0032] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0033] It should be understood that when the terms "first", "second", etc. are used in the claims, descriptions, and drawings of the present invention, they are only used to distinguish different objects, rather than to describe a specific order. The terms "include" and "comprise" used in the description and claims of the present invention indicate the presence of the described features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their collections.

[0034] It should be noted that in order to achieve comprehensive control of pressure and temperature during the single crystal growth process and reduce the generation of crystal twins, the pressure of the high-pressure furnace needs to be controlled in real time. Although the existing technology can achieve the control of equipment pressure to a certain extent, in the process of implementing proportional-integral-derivative (PID) control, there may be noise interference in the pressure data input to PID, resulting in the inability to accurately adjust the pressure value of the equipment.

[0035] Based on this, an embodiment of the present invention discloses a method for intelligently controlling the pressure in a crystal furnace. The method obtains the noise level of each pressure data point in a pressure data sequence, corrects the weight according to the noise level of the pressure data point, and can accurately identify the noise data in the pressure data sequence, thereby accurately realizing intelligent control of the pressure in the crystal furnace.

[0036] See also Figure 1 As shown, Figure 1 A flow chart of a method for intelligently controlling pressure in a crystal furnace provided in an embodiment of the present invention, the method specifically comprises the following steps S1 to S4.

[0037] S1: Obtain a preset local range of each pressure data point in the pressure data sequence.

[0038] For example, in an embodiment of the present invention, a preset local range of each pressure data point in a pressure data sequence is obtained, which also includes: collecting the pressure value in the crystal furnace through a pressure sensor, taking each pressure value as a pressure data point, and forming a pressure data sequence of the crystal furnace after preprocessing.

[0039] Among them, the acquisition frequency of the pressure sensor can be set to 5 Hz, and the acquisition frequency can be set according to actual needs; the preprocessing method can be set according to actual needs.

[0040] It should be noted that the pressure sensor may be affected by electromagnetic interference, electronic noise, etc. during the process of collecting pressure data, resulting in noise in the pressure data. The fluctuation of noise data in the pressure data of the crystal furnace will affect the accuracy of the PID algorithm, resulting in the inability to accurately control the pressure in the crystal furnace.

[0041] Based on this, an embodiment of the present invention identifies noise points in a pressure data sequence by using a clustering outlier detection algorithm based on a minimum spanning tree (MST). The algorithm constructs a minimum spanning tree of a data set and uses the structure of the tree to calculate the similarity between data points to identify outliers that are significantly different from most data points. These outliers are the noise points in the data set.

[0042] For example, in the embodiment of the present invention, the number of extreme value points in the local range can be obtained by using the difference method.

[0043] Among them, the size of the local range can be 21, and the size of the local range can be set according to actual needs; the size of the local range is the number of data points contained in the local range.

[0044] After obtaining each pressure data point in the pressure data sequence based on the above steps, continue to perform the following steps.

[0045] S2: Get the number of extreme value points in the local range and calculate the noise performance of the pressure data points in the pressure data sequence.

[0046] It should be noted that the traditional MST clustering-based outlier detection algorithm separates outliers by calculating the similarity between data points as weights. However, during the use of the crystal furnace, the data collection process may be disturbed by the environment, resulting in noise fluctuations.

[0047] Based on this, the embodiment of the present invention obtains the noise performance degree of the pressure data points by acquiring the fluctuation intensity and randomness of each pressure data point in the pressure data sequence, and corrects the weights between the pressure data points based on the noise performance degree of the pressure data points, so that the outlier detection algorithm based on MST clustering can accurately separate the outliers.

[0048] For example, when obtaining the noise performance of a pressure data point, it can be characterized by obtaining data fluctuations in a local range of the pressure data point. The greater the data fluctuation, the greater the possibility that the pressure data point is noise data.

[0049] Specifically, when obtaining the data fluctuation in the pressure data sequence, the extreme value points in the local range of the pressure data points can be obtained. The more irregular the distribution of the extreme value points in the local range of the pressure data points, the greater the data fluctuation. If the number of extreme value points in the local range is less than 3, it means that the overall data fluctuation in the local range is small, the data trend is relatively stable, and the noise performance degree of the pressure data point is 0; if the number of extreme value points in the local range is greater than or equal to 3, it means that the overall data fluctuation intensity in the local range is large. In order to accurately obtain the consistency of the specific distribution of the extreme value points and thus determine the noise performance degree of the pressure data points, the Euclidean distance between adjacent extreme value points can also be obtained. The larger the variance of the Euclidean distance between adjacent extreme value points in the local range of the pressure data points, the worse the overall consistency of the data, and the higher the noise level of the pressure data points.

[0050] For example, in the embodiment of the present invention, the noise performance degree of the pressure data point is determined, and the following relationship can be specifically referred to:

[0051] ;

[0052] In the formula, Indicates the noise performance of the i-th pressure data point, represents the variance function, represents the variance of the Euclidean distances between all adjacent extreme points in the local range of the i-th pressure data point, represents the number of extreme points in the local range of the i-th pressure data point, , Respectively represent the jth and jth pressure data points in the local range of the i-th pressure data point The pressure value of the extreme point, Indicates the absolute value symbol;

[0053] In the above formula, It can characterize the randomness of data fluctuations in the local range of pressure data points. Indicates the local range of the i-th pressure data point The absolute value of the difference between the pressure value of the extreme point and the pressure value of the jth extreme point, It represents the mean of the pressure deviations between all adjacent extreme points. The larger the mean, the greater the data fluctuation intensity in the local range, and the higher the noise performance of the i-th pressure data point.

[0054] In this way, the embodiment of the present invention can accurately obtain the noise performance degree of the pressure data point by specifically analyzing the data changes in the local range of the pressure data point, and then continue to execute the following steps.

[0055] S3: The distance between the pressure data points is used as a weight, and the weight is corrected according to the noise performance of the pressure data points.

[0056] It should be noted that after obtaining the noise performance degree of the pressure data points based on the above steps, the weights between the pressure data points can be corrected. However, the embodiment of the present invention also takes into account that the chemical reaction and the injection of nitrogen during the operation of the crystal furnace may cause random fluctuations in the normal and stable pressure data. If the noise data in the pressure data sequence is obtained by weight correction directly based on the noise performance degree obtained in the above steps, such normal data fluctuations may be mistakenly classified as noise data, thereby affecting the accuracy of the crystal furnace pressure control.

[0057] Based on this, the embodiment of the present invention further obtains parameters in the crystal furnace that are correlated with the pressure data, first corrects the noise performance level obtained based on the pressure data sequence, and then uses the corrected noise performance level to correct the weights between the pressure data points to distinguish between normal data fluctuations and noise data.

[0058] It is understandable that when a chemical reaction occurs in the crystal furnace or nitrogen is injected, it will not only cause changes in the pressure data but also changes in the temperature. From the ideal gas law, it can be known that when the volume and molar number in a closed space remain unchanged, the pressure and temperature are positively correlated. Therefore, the correlation between the temperature data and the pressure data in the same time period can be used to correct the degree of noise performance. Among them, the stronger the positive correlation between the temperature data and the pressure data, the more likely it is that the pressure data is a normal change, and the lower its noise performance is.

[0059] By way of example, in an embodiment of the present invention, the weight is corrected by the noise performance degree of the pressure data point, and the method also includes: obtaining a temperature data sequence of the crystal furnace, obtaining a temperature data segment corresponding to a local range in the temperature data sequence, and determining the Pearson correlation coefficient between the pressure data segment and the temperature data segment corresponding to the local range; after correcting the noise performance degree by the Pearson correlation coefficient, the possible noise degree of the pressure data point is obtained; and the product of the absolute value of the difference between the possible noise degrees of any two pressure data points and the weight is used as the correction weight between the two pressure data points.

[0060] Among them, the pressure value and temperature value at the same collection time correspond one by one.

[0061] For example, the closer the value of the Pearson correlation coefficient between the pressure data and the temperature data is to 1, the stronger the positive correlation between the pressure data and the temperature data is; determining the Pearson correlation coefficient between the pressure data segment and the temperature data segment can be achieved through existing technology, and the embodiments of the present invention will not be described in detail here.

[0062] For example, in the embodiment of the present invention, the possible degree of noise of the pressure data point is determined, and the specific details can be referred to the following relationship:

[0063] ;

[0064] In the formula, represents the possible degree of noise of the ith pressure data point, Indicates the noise performance of the i-th pressure data point, represents the pressure data segment corresponding to the local range of the i-th pressure data point, Represents the temperature data segment corresponding to the local range of the i-th pressure data point, represents the Pearson correlation coefficient between the pressure data segment and the temperature data segment corresponding to the local range of the i-th pressure data point, Represents the normalization function.

[0065] In the above formula, the larger the Pearson correlation coefficient between the pressure data segment and the temperature data segment corresponding to the local range, the stronger the positive correlation between the pressure data and the temperature data, the stronger the correction of the noise performance degree of the i-th pressure data point, and the lower the noise probability degree of the i-th pressure data point. The Pearson correlation coefficient between the pressure data segment and the temperature data segment corresponding to the local range of the pressure data point is negatively correlated with the noise probability degree of the pressure data point.

[0066] After the noise probability of the pressure data points is obtained based on the above steps, the weights between the pressure data points can be corrected according to the noise probability between the pressure data points.

[0067] For example, the distance between the pressure data points may be obtained by Euclidean distance, Pearson correlation coefficient, Minkowski distance, etc., which may be specifically set according to actual needs and will not be elaborated in detail in the embodiment of the present invention.

[0068] After the weights are corrected according to the possible degree of noise between the pressure data points obtained in the above steps, outlier detection can be performed based on the corrected weights to identify noise points in the pressure data sequence.

[0069] S4: Using the correction weights between the pressure data points in the MST clustering outlier detection algorithm to denoise the pressure data sequence, the denoised pressure data sequence is input into the PID algorithm to achieve pressure control in the crystal furnace.

[0070] It should be noted that PID control based on the denoised pressure data sequence can effectively avoid the interference of noise data, thereby improving the accuracy of pressure control inside the crystal furnace.

[0071] By way of example, in an embodiment of the present invention, the pressure data sequence is denoised using the correction weights between pressure data points in an MST clustering outlier detection algorithm, including: taking each pressure data point in the pressure data sequence as a node, and constructing a weighted graph using the correction weights between nodes as edges; obtaining the number of edges between the nodes of the minimum spanning tree by acquiring the minimum spanning tree in the weighted graph; dividing the minimum spanning tree into multiple spanning subtrees by cutting off the M edges with the largest correction weights, where M is a preset number of cut edges; taking nodes in the spanning subtrees whose correction weights between nodes exceed the preset weights as outliers, and eliminating them to obtain a denoised pressure data sequence.

[0072] Among them, the number of cut edges can be set according to actual needs, and the embodiment of the present invention does not impose too many restrictions on this; the preset weights can be set according to actual needs, and the embodiment of the present invention does not impose too many restrictions on this; the specific steps of obtaining the minimum spanning tree in the weighted graph and obtaining the number of edges between the minimum spanning tree nodes can be implemented by the Prim algorithm, and the embodiment of the present invention will not be repeated here.

[0073] It can be understood that the larger the correction weight between the nodes in the generated subtree, the greater the possibility that the corresponding two nodes are outliers. Therefore, if the correction weight between one of the nodes in the generated subtree and all other nodes exceeds the preset weight, the node is an outlier and the pressure data point corresponding to the node is noise data.

[0074] It should be further explained that in the MST clustering results, outlier data can only fall into the minimum spanning subtree with a smaller number of nodes. Therefore, when detecting outliers, only the spanning subtree with a smaller number of nodes needs to be detected. This can effectively reduce the number and time of data scanning and improve the efficiency of outlier detection. The specific settings can be made according to actual needs, and the embodiments of the present invention will not be described in detail here.

[0075] In this way, the embodiment of the present invention can effectively identify and remove noise points in the data by performing denoising processing based on the MST clustering outlier detection algorithm. In the process of constructing the MST, the algorithm can retain important information of the data sequence while denoising, thereby more accurately reflecting the changing trend and fluctuation of the pressure and providing more reliable input data for the PID algorithm.

[0076] For example, in an embodiment of the present invention, nodes in the generated subtree whose corrected weights exceed preset weights are regarded as outliers and are removed. Interpolation method can also be used to insert new pressure values ​​at the positions of the noise data to obtain a complete denoised pressure data sequence.

[0077] Based on the above steps, a denoised pressure data sequence is obtained, through which the pressure in the crystal furnace can be precisely controlled.

[0078] By way of example, in an embodiment of the present invention, the denoised pressure data sequence is input into a PID algorithm to achieve pressure control in a crystal furnace, including: obtaining a target pressure value required for the crystal in the crystal furnace; inputting the denoised pressure data sequence into a PID algorithm, and the PID algorithm determines the nitrogen input amount in the crystal furnace through current pressure deviation, integral accumulation and integral prediction.

[0079] The current pressure deviation is the deviation between the pressure value of the pressure data point and the target pressure value. The larger the deviation, the larger the nitrogen adjustment amount is, so as to quickly reduce the deviation.

[0080] Integration eliminates static errors by analyzing the accumulation of pressure deviations over a historical period of time. Integration helps to gradually adjust the nitrogen injection volume over a long period of time to maintain a stable pressure balance.

[0081] Integral prediction predicts the future pressure change trend and adjusts the nitrogen injection amount in advance according to the prediction results. The differential term helps to suppress overshoot and oscillation of pressure and improve stability and response speed.

[0082] In this way, the embodiment of the present invention can accurately control the pressure during the crystal growth process by processing the denoised pressure data sequence through PID control, thereby reducing the possibility of crystal twinning.

[0083] After the crystal furnace pressure control is realized based on the above steps, the embodiment of the present invention also considers that although the PID controller is powerful, it may also be affected by external interference, equipment failure or improper parameter setting, resulting in unsatisfactory or failure control effect. Through abnormal monitoring, these problems can be discovered in time and corresponding measures can be taken.

[0084] For example, in an embodiment of the present invention, the pressure in the crystal furnace is controlled, and then the method further includes: performing abnormal monitoring on the denoised pressure data sequence and issuing an early warning of the abnormal monitoring result.

[0085] In this way, the embodiment of the present invention can timely discover abnormal fluctuations and avoid the occurrence of potential dangers by issuing an abnormal pressure warning to the outside.

[0086] It can be seen that in an embodiment of the present invention, when realizing the control of the pressure in the crystal furnace, the preset local range of each pressure data point in the pressure data sequence can be obtained to obtain the number of extreme points in the local range; the noise performance degree of the pressure data points in the pressure data sequence is calculated; the distance between the pressure data points is used as a weight, and the weight is corrected according to the noise performance degree of the pressure data points; the pressure data sequence is denoised using the corrected weights between the pressure data points in the MST clustering outlier detection algorithm, and the denoised pressure data sequence is input into the PID algorithm to realize the control of the pressure in the crystal furnace.

[0087] In this way, the embodiment of the present invention corrects the weights between pressure data points by obtaining the noise performance degree of the pressure data points, thereby effectively avoiding the influence of noise data fluctuations on weight calculation and improving the accuracy of crystal furnace pressure control; on this basis, the embodiment of the present invention also obtains temperature data that is correlated with the pressure data, and corrects the noise performance degree of the pressure data points by using the correlation between the pressure data and the temperature data, thereby effectively avoiding the influence of normal data fluctuations on weight correction, thereby effectively improving the accuracy of crystal furnace pressure control.

[0088] An embodiment of the present invention further discloses an intelligent control system for pressure inside a crystal furnace, comprising a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, an intelligent control method for pressure inside a crystal furnace according to the present invention is implemented.

[0089] The above system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface, and their configuration and functions are known in the art, so they will not be described in detail here.

[0090] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium may be any suitable magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM, a dynamic random access memory DRAM, a static random access memory SRAM, an enhanced dynamic random access memory EDRAM, a high bandwidth memory HBM, a hybrid memory cube HMC, etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of a device or accessible or connectable to a device.

[0091] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

[0092] The above are all preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for intelligently controlling pressure in a crystal furnace, characterized in that: include: Obtain a preset local range of each pressure data point in the pressure data sequence, and obtain the number of extreme value points in the local range; calculate the noise performance degree of the pressure data point in the pressure data sequence: ； In the formula, Indicates the noise performance of the i-th pressure data point, represents the variance function, represents the variance of the Euclidean distances between all adjacent extreme points in the local range of the i-th pressure data point, represents the number of extreme points in the local range of the i-th pressure data point, , Respectively represent the jth and jth pressure data points in the local range of the i-th pressure data point The pressure value of the extreme point, Indicates the absolute value symbol; The distance between the pressure data points is used as a weight, and the weight is corrected according to the noise performance of the pressure data points, including: Acquire a temperature data sequence of a crystal furnace, acquire a temperature data segment corresponding to the local range in the temperature data sequence, and determine a Pearson correlation coefficient between a pressure data segment and a temperature data segment corresponding to the local range; After the noise performance degree is corrected by the Pearson correlation coefficient, the possible noise degree of the pressure data point is obtained, which satisfies the relationship: ； In the formula, represents the possible degree of noise of the ith pressure data point, represents the pressure data segment corresponding to the local range of the i-th pressure data point, Represents the temperature data segment corresponding to the local range of the i-th pressure data point, represents the Pearson correlation coefficient between the pressure data segment and the temperature data segment corresponding to the local range of the i-th pressure data point, represents the normalization function; The product of the absolute value of the difference in the possible degree of noise between any two pressure data points and the weight is used as the correction weight between the two pressure data points; The correction weights between the pressure data points are used in the MST clustering outlier detection algorithm to denoise the pressure data sequence, and the denoised pressure data sequence is input into the PID algorithm to achieve pressure control in the crystal furnace.

2. The method for intelligently controlling pressure in a crystal furnace according to claim 1, characterized in that: The step of obtaining a preset local range of each pressure data point in the pressure data sequence further includes: The pressure value in the crystal furnace is collected by the pressure sensor, and each pressure value is taken as a pressure data point, which constitutes the pressure data sequence of the crystal furnace after preprocessing.

3. The method for intelligently controlling pressure in a crystal furnace according to claim 1, characterized in that: The method of using the correction weights between the pressure data points in the MST clustering outlier detection algorithm to denoise the pressure data sequence comprises: Taking each pressure data point in the pressure data sequence as a node and taking the correction weights between the nodes as edges to construct a weighted graph; obtaining the number of edges between the nodes of the minimum spanning tree by obtaining the minimum spanning tree in the weighted graph; The minimum spanning tree is divided into multiple spanning subtrees by cutting off the M edges with the largest correction weights, where M is the preset number of cut edges; nodes in the spanning subtrees whose correction weights between nodes exceed the preset weights are regarded as outliers and are eliminated to obtain a denoised pressure data sequence.

4. The method for intelligently controlling pressure in a crystal furnace according to claim 1, characterized in that: The de-noised pressure data sequence is input into a PID algorithm to control the pressure in the crystal furnace, including: Obtaining a target pressure value required for the crystal in the crystal furnace; The denoised pressure data sequence is input into the PID algorithm, and the PID algorithm determines the nitrogen input amount in the crystal furnace through the current pressure deviation, integral accumulation and integral prediction.

5. The method for intelligently controlling pressure in a crystal furnace according to claim 1, characterized in that: The method of achieving the control of the pressure in the crystal furnace further includes: The denoised pressure data sequence is monitored for abnormalities, and an early warning is issued for abnormal monitoring results.

6. An intelligent control system for pressure in a crystal furnace, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a method for intelligently controlling pressure in a crystal furnace according to any one of claims 1 to 5 is implemented.

Images