A linkage regulation method and system for a vacuum environment

Abstract

The application discloses a kind of linkage regulation and control methods, system of vacuum environment, belong to vacuum environment regulation and control technical field, it specifically includes: real-time acquisition crystal growth environment time series data, constructs temperature-vacuum-growth phase three-dimensional mapping relationship, dynamic correction vacuum threshold value is combined furnace body material thermal expansion deformation compensation algorithm, generates multi-parameter coupling vacuum control model;Based on vacuum control model generates the dynamic vacuum threshold curve of each growth phase, through real-time deviation calculation output cooperative control signal;Three-level linkage control strategy is used to maintain vacuum degree stability, and the change trend of vacuum degree is predicted using long short-term memory network, and the standby vacuum unit is activated in advance;Through distributed fiber optic strain sensor detection sealing surface leakage acoustic characteristic, locate failure point and drive magnetic compensation plug dynamic plugging, simultaneously feedback leakage information to vacuum control model, realize closed-loop optimization;The application improves the vacuum degree control precision and equipment reliability of crystal growth process.

Description

Technical Field

[0001] This invention belongs to the field of vacuum environment control technology, specifically a method and system for the coordinated control of vacuum environment. Background Technology

[0002] In the growth of high-purity germanium single crystals, the stability of the vacuum environment is crucial to crystal quality. Traditional control methods treat vacuum level control, impurity filtration, temperature regulation, and gas flow management as relatively independent processes, lacking synergistic coordination between parameters. This approach struggles to adjust the vacuum environment promptly and accurately in response to the complex changes during the actual growth process. Consequently, germanium single crystals are susceptible to the influence of impurity gases, leading to crystal defects and affecting the uniformity and purity of crystal growth. For example, sudden gas flow changes during pump switching can cause vacuum chamber vibrations, blockages in the impurity filtration unit can reduce exhaust gas treatment efficiency, and uneven temperature fields can cause unstable crystal growth rates. Furthermore, existing control methods have low automation levels, relying on manual judgment and operation, resulting in low efficiency and failing to meet the demands of large-scale, high-quality germanium single crystal production. Therefore, a multi-dimensional, synergistic vacuum environment control method is urgently needed to improve overall control efficiency and crystal growth quality. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention proposes a method and system for the coordinated control of the vacuum environment. It collects real-time time-series data of the crystal growth environment, constructs a three-dimensional mapping relationship between temperature, vacuum, and growth stages, and dynamically corrects the vacuum threshold using a furnace material thermal expansion deformation compensation algorithm, generating a multi-parameter coupled vacuum control model. Based on the vacuum control model, it generates dynamic vacuum threshold curves for each growth stage and outputs coordinated control signals through real-time deviation calculation. A three-level coordinated control strategy is employed to maintain vacuum stability, and a long short-term memory network is used to predict vacuum change trends and activate backup vacuum units in advance. Distributed fiber optic strain sensors detect leakage acoustic characteristics at the sealing surface, locate failure points, and drive a magnetic compensation plug for dynamic sealing. Simultaneously, leakage information is fed back to the vacuum control model to achieve closed-loop optimization. This invention improves the vacuum control accuracy and equipment reliability during the crystal growth process.

[0004] To achieve the above objectives, the present invention provides the following technical solution:

[0005] A method for coordinated control of a vacuum environment, comprising:

[0006] S1: Real-time acquisition of crystal growth environment time-series data, establishment of a three-dimensional mapping relationship between temperature, vacuum, and growth stage, combined with a thermal expansion deformation compensation algorithm, dynamically correcting the preset vacuum threshold according to the thermal expansion coefficient of the furnace material, and generating a multi-parameter coupled vacuum control model; the crystal growth environment time-series data includes temperature gradient distribution, gas component concentration, and vacuum degree time-series fluctuation data in the growth furnace.

[0007] S2: Based on the vacuum control model, a dynamic vacuum threshold curve is generated for each crystal growth stage. According to the deviation between the current vacuum environment parameters and the dynamic vacuum threshold curve obtained in real time, a collaborative control signal is calculated and output synchronously. The collaborative control signal includes vacuum pump group start / stop command, micro-leakage compensation command, and inert gas injection command.

[0008] S3: According to the cooperative control signal, the vacuum fluctuation range inside the furnace is controlled within the preset range through a three-level linkage control method, and the vacuum change trend is predicted by using a long short-term memory network. When the prediction deviation exceeds the set vacuum change trend threshold, the backup vacuum unit is activated in advance.

[0009] S4: During the vacuum control process in S3, distributed fiber optic strain sensors arranged on the furnace body sealing surface collect the strain data of the sealing surface. The leakage acoustic fingerprint recognition algorithm is used to analyze the acoustic characteristics in the sealing surface strain data and locate the failure point. Based on the location result of the failure point, the magnetic compensation plug is automatically controlled to move to the coordinates of the failure point to implement dynamic sealing. At the same time, the leakage detection and sealing information is fed back to the vacuum control model.

[0010] Specifically, the specific steps of S1 include:

[0011] S1.1: Real-time data of the crystal growth environment inside the growth furnace is collected by sensors deployed inside the furnace and preprocessed; the preprocessing includes cleaning and normalization.

[0012] S1.2: Collect temperature, vacuum level, and growth stage data during historical crystal growth processes, and integrate them with preprocessed crystal growth environment time-series data. Establish a three-dimensional mapping relationship between temperature, vacuum, and growth stage using a support vector machine. The growth stage data is obtained by recording the time nodes and growth stage characteristics of crystal growth.

[0013] S1.3: Establish a database of thermal expansion coefficient-temperature relationship by experimentally measuring the thermal expansion coefficient of the furnace material at different temperatures;

[0014] S1.4: Based on the real-time collected temperature gradient distribution data, the furnace body is divided into N local regions. For each local region, the corresponding thermal expansion coefficient is queried from the thermal expansion coefficient-temperature relationship database according to the temperature value of the local region. The temperature deviation is obtained by calculating the difference between the current temperature and the initial temperature of the local region. Then, the expansion amount of the local region is obtained by multiplying the temperature deviation by the length of the local region and the corresponding thermal expansion coefficient obtained from the query.

[0015] S1.5: By calculating the expansion amount of each local region, the local deformation distribution of the furnace body is obtained.

[0016] Specifically, the specific steps of S1 further include:

[0017] S1.6: Based on the structural design drawing of the furnace body, determine the sealing gap under static working conditions. Based on the expansion of each local area and combined with the mechanical properties of the furnace body structure, construct a deformation-sealing gap correction model. Then, substitute the local deformation distribution of the furnace body into the deformation-sealing gap correction model to obtain the corrected effective working gap of the vacuum sealing surface.

[0018] S1.7: Based on the corrected effective working gap of the vacuum sealing surface, combined with the performance parameters of the vacuum system and the crystal growth process requirements, the dynamically corrected vacuum threshold is obtained;

[0019] S1.8: A multilayer perceptron neural network is used to model the preprocessed crystal growth environment time-series data, the three-dimensional mapping relationship between temperature, vacuum and growth stages, and the dynamically corrected vacuum threshold, generating a multi-parameter coupled vacuum control model.

[0020] Specifically, the steps of S2 include:

[0021] S2.1: Obtain the preprocessed crystal growth environment time series data and extract the characteristic parameters of the crystal growth stage; the crystal growth stage includes the initial stage, the stable growth stage and the final stage; the characteristic parameters of the crystal growth stage include the temperature change rate, the vacuum stability, and the crystal growth time nodes;

[0022] S2.2: Input the characteristic parameters of the crystal growth stage into the vacuum control model, and the vacuum control model outputs the dynamic vacuum threshold curve of the current growth stage;

[0023] S2.3: Real-time acquisition of current crystal growth environment time-series data through sensors, and calculation of the deviation between current vacuum degree time-series fluctuation data and dynamic vacuum threshold curve;

[0024] S2.4: Determine the start / stop status of the vacuum pump unit based on the magnitude and direction of the deviation;

[0025] If the deviation exceeds the preset deviation threshold, the vacuum pump unit will be started.

[0026] If the deviation is less than the preset deviation threshold, the vacuum pump unit will be stopped.

[0027] If the deviation is equal to the preset deviation threshold, the current state of the vacuum pump group remains unchanged.

[0028] S2.5: Calculate the micro-leakage compensation amount based on the magnitude of the deviation; the micro-leakage compensation amount is the product of the micro-leakage compensation coefficient and the deviation;

[0029] If the microleakage compensation is greater than zero, then increase the microleakage amount;

[0030] If the microleakage compensation is less than zero, then reduce the microleakage amount;

[0031] If the micro-leakage compensation amount is equal to zero, then the current micro-leakage compensation valve state remains unchanged;

[0032] S2.6: Calculate the inert gas injection amount based on the deviation and the current gas component concentration; the inert gas injection amount is the difference between the target gas component concentration and the current gas component concentration, multiplied by the inert gas injection coefficient and the deviation;

[0033] If the amount of inert gas injected is greater than zero, then inert gas is injected.

[0034] If the amount of inert gas injected is less than zero, then reduce the amount of inert gas injected.

[0035] If the inert gas injection amount is zero, then maintain the current state of the inert gas injection valve;

[0036] S2.7: Integrate the vacuum pump group start / stop command, micro-leakage compensation command, and inert gas injection command, and output the integrated coordinated control signal to the actuator, including the vacuum pump group controller, micro-leakage compensation valve, and inert gas injection valve.

[0037] Specifically, the three-level linkage control method includes:

[0038] The first level of control is the coarse adjustment control of the vacuum pump group. Based on the start and stop commands of the vacuum pump group in the coordinated control signal, the number of vacuum pump groups that are turned on and off is controlled to adjust the vacuum level in the furnace.

[0039] The second level of control is the fine-tuning control of the micro-leakage compensation valve. According to the micro-leakage compensation command, the flow rate of the micro-leakage compensation valve is adjusted to fine-tune the vacuum level inside the furnace.

[0040] The third level of control is the auxiliary control of the inert gas injection valve. Based on the inert gas injection command, it controls the injection amount and injection speed of the inert gas to further stabilize the vacuum level inside the furnace.

[0041] Specifically, the step of activating the backup vacuum unit in advance when the prediction deviation exceeds the set threshold for the vacuum degree change trend includes:

[0042] A threshold for the trend of vacuum degree change is set; the threshold for the trend of vacuum degree change is determined based on the crystal growth process requirements and equipment performance.

[0043] The vacuum degree change trend predicted by the Long Short-Term Memory Network is compared with the set vacuum degree change trend threshold.

[0044] If the prediction deviation is greater than the threshold of vacuum degree change trend, an activation command is sent to the backup vacuum unit to start the backup vacuum unit in advance and increase the pumping capacity of the vacuum system.

[0045] If the prediction deviation is less than the negative of the vacuum degree change trend threshold, an activation command is sent to the backup vacuum unit to start the backup vacuum unit in advance, thereby reducing the pumping capacity of the vacuum system or increasing the injection of inert gas.

[0046] If the prediction deviation equals the threshold of vacuum degree change trend, then the current control state remains unchanged.

[0047] Specifically, the specific steps of S4 include:

[0048] S4.1: Arrange M distributed fiber optic strain sensors around the sealing surface of the furnace body, determine the spatial coordinates of each sensor, and establish the coordinate system of the sealing surface;

[0049] S4.2: Establish a control system for the magnetic compensation plug, and simultaneously establish a mapping relationship between the sealing surface coordinate system and the robot arm motion coordinates; the control system for the magnetic compensation plug includes a drive motor, a transmission mechanism, and a position sensor;

[0050] S4.3: When the furnace body sealing surface is under normal operating conditions, the sealing surface strain data within a time interval ΔT is continuously collected using distributed fiber optic strain sensors and preprocessed to obtain preprocessed normal operating condition data; the normal operating condition is no leakage.

[0051] S4.4: For the preprocessed normal operating condition data, Fourier transform is used to extract the acoustic wave features and store them in the normal operating condition acoustic wave feature library of the sealing surface.

[0052] S4.5: Using distributed fiber optic strain sensors arranged on the sealing surface of the furnace body, the strain data of the sealing surface is collected in real time and preprocessed to obtain the preprocessed strain data.

[0053] S4.6: Wavelet packet transform is used to perform multi-scale analysis on the preprocessed strain data, calculate the energy of each frequency band, and compare it with the frequency band energy in the acoustic characteristic library of the sealing surface under normal working conditions.

[0054] When the energy of any frequency band deviates from the normal range, it is determined that the frequency band contains abnormal acoustic signals, and the leakage location process is initiated; the abnormal acoustic signal refers to an acoustic leakage signal.

[0055] Specifically, the steps of S4 further include:

[0056] S4.7: For the extracted abnormal acoustic wave signal, calculate its cross-correlation function between the signals received by different sensors, and determine the time difference of the acoustic wave arriving at different sensors by finding the peak position of the cross-correlation function.

[0057] S4.8: Based on the time difference of sound waves arriving at different sensors, and combined with the spatial coordinates of each sensor, a set of positioning equations is constructed. By solving the set of positioning equations, the three-dimensional coordinates of the failure point are calculated.

[0058] S4.9: Utilize the mapping relationship between the sealing surface coordinate system and the robotic arm motion coordinate system to convert the three-dimensional coordinates of the failure point into control coordinates in the robotic arm motion coordinate system;

[0059] S4.10: Based on the control coordinates in the motion coordinate system of the robotic arm, control commands are sent to the drive motor to control the drive motor to rotate according to the predetermined motion trajectory and speed, drive the magnetic compensation plug to move through the transmission mechanism, and use a path planning algorithm to plan the movement path during the movement of the magnetic compensation plug.

[0060] S4.11: The position information of the magnetic compensator is fed back in real time by the position sensor and compared with the target position in the control command. The control system adjusts the movement of the drive motor in real time according to the comparison result.

[0061] Specifically, the steps of S4 further include:

[0062] S4.12: When the magnetic compensating plug reaches the three-dimensional coordinates of the failure point, the control system controls the magnetic compensating plug to generate a magnetic field, and makes the magnetic compensating plug fit tightly against the furnace sealing surface under the action of the magnetic field, filling the leakage point of the sealing surface and completing the dynamic sealing of the leakage point.

[0063] S4.13: Verify the sealing effect by monitoring changes in the vacuum level inside the furnace;

[0064] If the vacuum level remains stable within the normal range, it indicates that the sealing was successful.

[0065] If the vacuum level is abnormal, adjust the position of the magnetic compensation plug or the magnetic field strength.

[0066] S4.14: Feedback the leakage detection and sealing information to the vacuum control model so that the vacuum control model can adjust the control strategy according to the actual situation and optimize the vacuum control effect; the leakage detection and sealing information includes leakage type, failure point location, and sealing effect.

[0067] A linkage control system for a vacuum environment includes: a data acquisition module, a collaborative control module, a trend prediction module, and a leak detection module;

[0068] The data acquisition module is used to acquire crystal growth environment data in real time and construct a multi-parameter coupled vacuum control model.

[0069] The collaborative control module generates a dynamic vacuum threshold curve based on the vacuum regulation model and calculates the collaborative control signal.

[0070] The trend prediction module maintains vacuum stability and predicts vacuum change trends through a three-level linkage control method.

[0071] The leakage detection module is used to detect leaks on the furnace body sealing surface, locate the failure point, and implement dynamic sealing.

[0072] Compared with the prior art, the beneficial effects of the present invention are:

[0073] 1. This invention proposes a linkage control system for a vacuum environment, and optimizes and improves its architecture, operation steps and processes. The system has the advantages of simple process, low investment and operating costs, and low production and working costs.

[0074] 2. This invention proposes a method for coordinated control of the vacuum environment. By establishing a three-dimensional mapping relationship between temperature, vacuum, and growth stages and combining it with a thermal expansion deformation compensation algorithm, the vacuum threshold is dynamically corrected, generating a multi-parameter coupled vacuum control model to achieve precise control of the vacuum level at each growth stage. The three-level coordinated control strategy combined with long short-term memory network prediction can effectively suppress vacuum level fluctuations and activate backup units in advance when the trend is abnormal, thereby improving the stability and reliability of the crystal growth process.

[0075] 3. This invention proposes a linkage control method for vacuum environment. By utilizing distributed fiber optic strain sensors and a leakage acoustic fingerprint recognition algorithm, leakage on the sealing surface can be detected in real time and the failure point can be located. Dynamic sealing is achieved through magnetic compensation plugs, avoiding manual intervention. The leakage detection and sealing information is fed back to the vacuum control model to form a closed-loop optimization mechanism, which further enhances the system's adaptability to complex working conditions, reduces equipment maintenance costs, and extends service life. Attached Figure Description

[0076] Figure 1 This is a schematic diagram of a linkage control method for a vacuum environment according to the present invention;

[0077] Figure 2 This is a flowchart illustrating the principle of a linkage control method for a vacuum environment according to the present invention.

[0078] Figure 3 This is a diagram illustrating the architecture of a linkage control system for a vacuum environment according to the present invention. Detailed Implementation

[0079] Example 1:

[0080] Please see Figure 1 and Figure 2 The present invention provides an embodiment of a method for the coordinated control of a vacuum environment, comprising the following steps:

[0081] S1: Real-time acquisition of temperature gradient distribution, gas component concentration, and vacuum degree fluctuation data in the growth furnace; establishment of a three-dimensional mapping relationship between temperature, vacuum, and growth stages; combined with a thermal expansion deformation compensation algorithm; dynamic correction of the preset vacuum threshold based on the thermal expansion coefficient of the furnace material; generation of a multi-parameter coupled vacuum control model.

[0082] Furthermore, the real-time acquisition of temperature gradient distribution data within the growth furnace specifically includes:

[0083] (1) Multiple high-precision temperature sensors are arranged at different heights and radial positions inside the growth furnace. The spacing between the temperature sensors is determined according to the size of the growth furnace and the crystal growth process requirements to ensure that the temperature information at each position inside the furnace can be accurately obtained.

[0084] (2) Data from each temperature sensor is collected in real time through the data acquisition system, and the collected temperature data is filtered to remove noise interference and obtain accurate temperature gradient distribution data.

[0085] Furthermore, the real-time acquisition of gas component concentration data within the growth furnace specifically includes:

[0086] (1) Install a gas component analyzer in the growth furnace. The gas component analyzer uses spectral analysis technology and can detect the concentration of various gas components in the furnace in real time.

[0087] (2) Set the sampling frequency of the gas component analyzer to ensure that changes in gas component concentration can be captured in a timely manner.

[0088] S2: Based on the vacuum control model, a dynamic vacuum threshold curve is generated for each crystal growth stage. According to the deviation between the current vacuum environment parameters and the dynamic vacuum threshold curve obtained in real time, a collaborative control signal is calculated and output synchronously. The collaborative control signal includes vacuum pump group start / stop command, micro-leakage compensation command, and inert gas injection command.

[0089] S3: According to the cooperative control signal, the vacuum fluctuation range inside the furnace is controlled within the preset range through a three-level linkage control method, and the vacuum change trend is predicted by using a long short-term memory network. When the prediction deviation exceeds the set vacuum change trend threshold, the backup vacuum unit is activated in advance.

[0090] S4: During the vacuum control process in S3, distributed fiber optic strain sensors arranged on the furnace body sealing surface collect the strain data of the sealing surface. The leakage acoustic fingerprint recognition algorithm is used to analyze the acoustic characteristics in the sealing surface strain data and locate the failure point. Based on the location result of the failure point, the magnetic compensation plug is automatically controlled to move to the coordinates of the failure point to implement dynamic sealing. At the same time, the leakage detection and sealing information is fed back to the vacuum control model.

[0091] The specific steps of S1 include:

[0092] S1.1: Real-time data of the crystal growth environment inside the growth furnace is collected by sensors deployed inside the furnace and preprocessed; the preprocessing includes cleaning and normalization.

[0093] S1.2: Collect temperature, vacuum level, and growth stage data during historical crystal growth processes, and integrate them with preprocessed crystal growth environment time-series data. Establish a three-dimensional mapping relationship between temperature, vacuum, and growth stage using a support vector machine. The growth stage data is obtained by recording the time nodes and growth stage characteristics of crystal growth.

[0094] Furthermore, the specific steps in S1.2 include:

[0095] (1) Extract historical temperature and vacuum time series data, as well as corresponding growth stage labels, from the crystal growth database;

[0096] (2) Extract growth stage features, including time nodes and stage features;

[0097] Time nodes: Record the start or end time of each growth stage;

[0098] Stage characteristics: Statistical characteristics are calculated from time series data, such as mean temperature, standard deviation of vacuum degree, and temperature gradient;

[0099] (3) Integrate historical temperature and vacuum time series data, as well as extracted time nodes and stage features, to form a three-dimensional dataset. Each row of data in the three-dimensional dataset contains a temperature value, vacuum time series data and corresponding growth stage label at a time point.

[0100] (4) Use the time series data of temperature and vacuum degree as input features and the growth stage as output label to train the support vector machine model;

[0101] (5) Draw the decision boundary on the PT plane to divide the regions of different growth stages;

[0102] (6) For any given time series data of temperature and vacuum degree, the corresponding growth stage prediction is output through the support vector machine decision function.

[0103] S1.3: Establish a database of thermal expansion coefficient-temperature relationship by experimentally measuring the thermal expansion coefficient of the furnace material at different temperatures;

[0104] It needs to be explained that the expansion of the furnace body after being heated will cause deformation of the sealing surface, which directly affects the vacuum sealing gap. If thermal expansion is ignored, the vacuum control will deviate due to sealing failure. For example, if the expansion causes the gap to increase, the actual vacuum level will be lower than the set value. Therefore, it is necessary to establish a mapping relationship between material properties and temperature to provide parameters for deformation calculation.

[0105] S1.4: Based on the real-time collected temperature gradient distribution data, the furnace body is divided into N local regions. For each local region, the corresponding thermal expansion coefficient is queried from the thermal expansion coefficient-temperature relationship database according to the temperature value of the local region. The temperature deviation is obtained by calculating the difference between the current temperature and the initial temperature of the local region. Then, the expansion amount of the local region is obtained by multiplying the temperature deviation by the length of the local region and the corresponding thermal expansion coefficient obtained from the query.

[0106] It should be explained that the uneven heating of the furnace body causes different expansion amounts in different parts. Therefore, the deformation of the sealing surface has spatial differences. By calculating the area, the area affected by the deformation on the sealing gap can be accurately located.

[0107] S1.5: By calculating the expansion amount of each local region, the local deformation distribution of the furnace body is obtained;

[0108] The local deformation distribution of the furnace body is the superposition of the expansion amount of each local region.

[0109] S1.6: Based on the structural design drawing of the furnace body, determine the sealing gap under static working conditions. Based on the expansion of each local area and combined with the mechanical properties of the furnace body structure, construct a deformation-sealing gap correction model. Then, substitute the local deformation distribution of the furnace body into the deformation-sealing gap correction model to obtain the corrected effective working gap of the vacuum sealing surface.

[0110] The structural design of the furnace body includes the flange sealing surface inclination angle and bolt distribution; the sealing gap refers to the physical distance between two sealing components in the vacuum system, such as the flange and the sealing ring. In crystal growth furnaces, the sealing gap usually appears at the furnace body interface, observation window, inlet and outlet, etc., which directly affects the airtightness and vacuum maintenance capability of the system.

[0111] It should be explained that the sealing gap directly affects the leakage rate of the vacuum system. For example, the larger the gap, the greater the leakage. If it is not corrected, the preset vacuum threshold will be out of touch with the actual needs. For example, if the gap increases, a higher pumping rate is required to maintain the vacuum. This invention uses a deformation-sealing gap correction model to convert the deformation into a quantitative change in the sealing gap, providing physical parameters for the dynamic adjustment of the vacuum threshold.

[0112] S1.7: Based on the corrected effective working gap of the vacuum sealing surface, combined with the performance parameters of the vacuum system and the crystal growth process requirements, the dynamically corrected vacuum threshold is obtained;

[0113] Furthermore, the specific steps in S1.7 include:

[0114] (1) Obtain relevant performance parameters from the control instruments or sensors of the vacuum system, such as the pumping speed of the vacuum pump, the flow conductance of the pipeline, and the leakage rate of the system. Evaluate the collected performance parameters to determine whether they meet the requirements of the crystal growth process. For example, if the pumping speed of the vacuum pump is too low, the system may not be able to reach the required vacuum level.

[0115] (2) Consult the crystal growth process specifications to understand the specific requirements for vacuum degree at different crystal growth stages. For example, a higher vacuum degree is required in the nucleation stage of crystal growth, while the vacuum degree can be appropriately reduced in the later stage of growth.

[0116] (3) Based on the results of the process survey, set the vacuum threshold range for each stage of crystal growth;

[0117] (4) Based on the effective working gap of the corrected vacuum sealing surface, the performance parameters of the vacuum system and the crystal growth process requirements, a calculation model for dynamically correcting the vacuum threshold is established.

[0118] (5) Substitute the real-time obtained corrected effective working gap of the sealing surface and vacuum system performance parameters into the calculation model to obtain the dynamically corrected vacuum threshold.

[0119] S1.8: A multilayer perceptron neural network is used to model the preprocessed crystal growth environment time series data, the three-dimensional mapping relationship between temperature, vacuum and growth stage, and the dynamically corrected vacuum threshold to generate a multi-parameter coupled vacuum control model. The modeling process of the multilayer perceptron neural network is existing technology in this field and is not an inventive solution of this application, so it will not be described in detail here.

[0120] The specific steps of S2 include:

[0121] S2.1: Obtain the preprocessed crystal growth environment time series data and extract the characteristic parameters of the crystal growth stage; the crystal growth stage includes the initial stage, the stable growth stage and the final stage; the characteristic parameters of the crystal growth stage include the temperature change rate, the vacuum stability, and the crystal growth time nodes;

[0122] S2.2: Input the characteristic parameters of the crystal growth stage into the vacuum control model, and the vacuum control model outputs the dynamic vacuum threshold curve of the current growth stage;

[0123] S2.3: Real-time acquisition of current crystal growth environment time-series data through sensors, and calculation of the deviation between current vacuum degree time-series fluctuation data and dynamic vacuum threshold curve;

[0124] S2.4: Determine the start / stop status of the vacuum pump unit based on the magnitude and direction of the deviation;

[0125] If the deviation is greater than the preset deviation threshold, it means that the vacuum is too high and the vacuum pump unit needs to be started.

[0126] If the deviation is less than the preset deviation threshold, it means that the vacuum is too low and the vacuum pump unit needs to be stopped.

[0127] If the deviation is equal to the preset deviation threshold, the current state of the vacuum pump group remains unchanged.

[0128] S2.5: Calculate the micro-leakage compensation amount based on the magnitude of the deviation; the micro-leakage compensation amount is the product of the micro-leakage compensation coefficient and the deviation, and the micro-leakage compensation coefficient is determined by those skilled in the art through a large number of experiments;

[0129] If the microleakage compensation is greater than zero, it means that the microleakage needs to be increased.

[0130] If the microleakage compensation is less than zero, it means that the microleakage needs to be reduced.

[0131] If the micro-leakage compensation amount is equal to zero, then the current micro-leakage compensation valve state remains unchanged;

[0132] S2.6: Calculate the inert gas injection amount based on the deviation and the current gas component concentration; the inert gas injection amount is the difference between the target gas component concentration and the current gas component concentration, multiplied by the inert gas injection coefficient and the deviation, and the inert gas injection coefficient is determined by those skilled in the art through a large number of experiments;

[0133] If the amount of inert gas injected is greater than zero, it means that inert gas needs to be injected.

[0134] If the amount of inert gas injected is less than zero, it means that the amount of inert gas injected needs to be reduced.

[0135] If the inert gas injection amount is zero, then maintain the current state of the inert gas injection valve;

[0136] S2.7: Integrate the vacuum pump group start / stop command, micro-leakage compensation command, and inert gas injection command, and output the integrated coordinated control signal to the actuator, including the vacuum pump group controller, micro-leakage compensation valve, and inert gas injection valve.

[0137] The three-level linkage control method specifically includes:

[0138] The first level of control is the coarse adjustment control of the vacuum pump group. Based on the start and stop commands of the vacuum pump group in the coordinated control signal, the number of vacuum pump groups that are turned on and off is controlled to quickly adjust the vacuum level in the furnace.

[0139] The second level of control is the fine-tuning control of the micro-leakage compensation valve. According to the micro-leakage compensation command, the flow rate of the micro-leakage compensation valve is adjusted to fine-tune the vacuum level inside the furnace.

[0140] The third level of control is the auxiliary control of the inert gas injection valve. Based on the inert gas injection command, it controls the injection amount and injection speed of the inert gas to further stabilize the vacuum level inside the furnace.

[0141] The step of activating the backup vacuum unit in advance when the prediction deviation exceeds the set threshold for vacuum degree change trend includes:

[0142] A threshold for the trend of vacuum degree change is set; the threshold for the trend of vacuum degree change is determined based on the crystal growth process requirements and equipment performance.

[0143] The vacuum degree change trend predicted by the Long Short-Term Memory Network is compared with the set vacuum degree change trend threshold.

[0144] If the prediction deviation is greater than the vacuum degree change trend threshold, an activation command is sent to the backup vacuum unit to start the backup vacuum unit in advance and enhance the pumping capacity of the vacuum system.

[0145] If the prediction deviation is less than the negative of the vacuum degree change trend threshold, an activation command is sent to the backup vacuum unit to start the backup vacuum unit in advance, thereby reducing the pumping capacity of the vacuum system or increasing the injection of inert gas.

[0146] If the prediction deviation equals the threshold of vacuum degree change trend, then the current control state remains unchanged.

[0147] The specific steps of S4 include:

[0148] S4.1: Arrange M distributed fiber optic strain sensors around the sealing surface of the furnace body, determine the spatial coordinates of each sensor, and establish the coordinate system of the sealing surface;

[0149] S4.2: Establish a control system for the magnetic compensation plug, and simultaneously establish a mapping relationship between the sealing surface coordinate system and the robot arm motion coordinates; the control system for the magnetic compensation plug includes a drive motor, a transmission mechanism, and a position sensor;

[0150] S4.3: When the furnace body sealing surface is under normal operating conditions, the sealing surface strain data within a time interval ΔT is continuously collected using distributed fiber optic strain sensors and preprocessed to obtain preprocessed normal operating condition data; the normal operating condition is no leakage.

[0151] S4.4: For the preprocessed normal operating condition data, Fourier transform is used to extract acoustic wave features and store them in the acoustic wave feature library of normal operating condition of sealing surface. The Fourier transform is the prior art in this field and is not an inventive solution of this application, so it will not be described in detail here.

[0152] S4.5: Using distributed fiber optic strain sensors arranged on the sealing surface of the furnace body, the strain data of the sealing surface is collected in real time and preprocessed to obtain the preprocessed strain data.

[0153] S4.6: Wavelet packet transform is used to perform multi-scale analysis on the preprocessed strain data, calculate the energy of each frequency band, and compare it with the frequency band energy in the acoustic wave feature library of the sealing surface under normal working conditions. Wavelet packet transform is the prior art in this field and is not an inventive solution of this application, so it will not be described in detail here.

[0154] When the energy of any frequency band deviates from the normal range, it is determined that the frequency band contains abnormal acoustic signals, and the leakage location process is initiated; the abnormal acoustic signal refers to an acoustic leakage signal.

[0155] S4.7: For the extracted abnormal acoustic wave signal, calculate its cross-correlation function between the signals received by different sensors, and determine the time difference of the acoustic wave arriving at different sensors by finding the peak position of the cross-correlation function.

[0156] Furthermore, the specific steps in S4.7 include:

[0157] (1) Obtain the abnormal acoustic signal collected by the furnace body sealing surface sensor;

[0158] (2) For each pair of sensors, calculate the cross-correlation function between the acoustic signals they receive. The cross-correlation function is used to measure the similarity between two abnormal acoustic signals under different time delays. It can be implemented by the similarity calculation formula. The similarity calculation formula is the prior art in this field and is not an inventive solution of this application. It will not be elaborated here.

[0159] (3) Find the peak position in the results of the cross-correlation function, where the peak position corresponds to the time delay required for the acoustic signal to propagate from one sensor to another.

[0160] (4) Calculate the time difference of sound waves arriving at different sensors based on the peak position.

[0161] S4.8: Based on the time difference of sound waves arriving at different sensors, and combined with the spatial coordinates of each sensor, a set of positioning equations is constructed. By solving the set of positioning equations, the three-dimensional coordinates of the failure point are calculated.

[0162] Furthermore, the specific steps of S4.8 include:

[0163] (1) Determine the spatial coordinates of each sensor, wherein the sensors are arranged in three-dimensional space, and the position of each sensor can be represented by coordinates in a three-dimensional spatial coordinate system;

[0164] (2) Obtain the time difference data of sound waves arriving at different sensors;

[0165] (3) Based on the speed of sound wave propagation, the time difference is converted into a distance difference; the speed of sound wave propagation refers to the propagation speed in air or a specific medium;

[0166] The conversion of time difference into distance difference is achieved by multiplying the time difference data by the speed of sound wave propagation;

[0167] (4) Construct a system of positioning equations using distance differences;

[0168] The positioning equations are expressed as follows: the distance difference is equal to the Euclidean distance difference between the spatial coordinates of each pair of sensors;

[0169] (5) Use the least squares method to solve for the three-dimensional coordinates of the failure point.

[0170] S4.9: Utilize the mapping relationship between the sealing surface coordinate system and the robotic arm motion coordinate system to convert the three-dimensional coordinates of the failure point into control coordinates in the robotic arm motion coordinate system;

[0171] S4.10: Based on the control coordinates in the motion coordinate system of the robotic arm, a control command is sent to the drive motor to control the drive motor to rotate according to the predetermined motion trajectory and speed. The magnetic compensation plug is moved through the transmission mechanism. During the movement of the magnetic compensation plug, a path planning algorithm is used to plan the movement path to avoid the high-temperature sensitive area inside the furnace. The path planning algorithm is the prior art in this field and is not an inventive solution of this application. It will not be described in detail here.

[0172] S4.11: The position information of the magnetic compensator is fed back in real time by the position sensor and compared with the target position in the control command. The control system adjusts the movement of the drive motor in real time according to the comparison result.

[0173] S4.12: When the magnetic compensating plug reaches the three-dimensional coordinates of the failure point, the control system controls the magnetic compensating plug to generate a magnetic field, and makes the magnetic compensating plug fit tightly against the furnace sealing surface under the action of the magnetic field, filling the leakage point of the sealing surface and completing the dynamic sealing of the leakage point.

[0174] S4.13: Verify the sealing effect by monitoring changes in the vacuum level inside the furnace;

[0175] If the vacuum level remains stable within the normal range, it indicates that the sealing was successful.

[0176] If the vacuum level is abnormal, further adjust the position of the magnetic compensation plug or the magnetic field strength.

[0177] S4.14: Feedback the leakage detection and sealing information to the vacuum control model so that the vacuum control model can adjust the control strategy according to the actual situation and optimize the vacuum control effect; the leakage detection and sealing information includes leakage type, failure point location, and sealing effect.

[0178] Example 2:

[0179] Please see Figure 3 Another embodiment of the present invention provides: a linkage control system for a vacuum environment, comprising:

[0180] Data acquisition module, collaborative control module, trend prediction module, and leak detection module;

[0181] The data acquisition module is used to collect crystal growth environment data in real time and construct a multi-parameter coupled vacuum control model;

[0182] The collaborative control module generates a dynamic vacuum threshold curve based on the vacuum regulation model and calculates the collaborative control signal.

[0183] The trend prediction module maintains vacuum stability and predicts vacuum change trends through a three-level linkage control method.

[0184] The leakage detection module is used to detect leaks on the furnace body sealing surface, locate the failure point, and implement dynamic sealing.

[0185] The data acquisition module includes: a data acquisition unit and a 3D mapping unit;

[0186] The data acquisition unit is used to collect data on temperature gradient distribution, gas component concentration, and vacuum fluctuation time series in the growth furnace.

[0187] The three-dimensional mapping unit is used to establish a three-dimensional mapping relationship between temperature, vacuum, and growth stages. Combined with the thermal expansion deformation compensation algorithm, the preset vacuum threshold is dynamically corrected according to the thermal expansion coefficient of the furnace material to generate a multi-parameter coupled vacuum control model.

[0188] The collaborative control module includes: a threshold generation unit and a signal generation unit;

[0189] Threshold generation unit, used to generate dynamic vacuum threshold curves for each crystal growth stage;

[0190] The signal generation unit is used to calculate the deviation between the current vacuum environment parameters and the dynamic vacuum threshold curve, and generate a coordinated control signal.

[0191] The trend prediction module includes: a three-level linkage control unit and a trend prediction unit;

[0192] The three-level linkage control unit controls the vacuum fluctuation range inside the furnace within a preset range through a three-level linkage method of PID adjustment, micro-leakage compensation, and inert gas injection, based on the coordinated control signal.

[0193] The trend prediction unit is used to predict the trend of vacuum level changes through a long short-term memory network. When the prediction deviation exceeds the set threshold, the backup vacuum unit is activated in advance.

[0194] The leak detection module includes: a leak detection unit, a dynamic sealing unit, and an information feedback unit;

[0195] The leakage detection unit is used to collect strain data of the sealing surface through distributed fiber optic strain sensors, analyze the acoustic characteristics using a leakage acoustic fingerprint recognition algorithm, and locate the failure point.

[0196] The dynamic plugging unit is used to automatically control the magnetic compensation plug to move to the coordinates of the failure point to implement dynamic plugging based on the failure point location result;

[0197] The information feedback unit is used to feed back leak detection and sealing information to the vacuum control model to achieve closed-loop optimization.

[0198] In summary, the data acquisition module provides the vacuum regulation model foundation for the dynamic threshold and collaborative control module. The output signal of the collaborative control module serves as the input of the trend prediction module, driving the three-level linkage control. The real-time data and prediction results of the trend prediction module can be fed back to the collaborative control module to optimize the control strategy. The leakage detection module monitors the sealing surface status in real time during the vacuum regulation process, and the sealing information is fed back to the data acquisition module to update the model parameters, forming a closed-loop control.

[0199] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments under the guidance of the present invention without departing from the spirit and scope of the present invention. All of these variations are within the protection scope of the present invention.

Claims

1. A linkage regulation method of a vacuum environment, characterized by, The application relates to a crystal growth environment vacuum control method and device. S1: real-time acquisition of crystal growth environment time series data, establishment of a temperature-vacuum-growth stage three-dimensional mapping relationship, combination of a thermal expansion deformation compensation algorithm, dynamic correction of a preset vacuum threshold value according to the thermal expansion coefficient of a furnace body material, generation of a vacuum control model coupled with multiple parameters; the crystal growth environment time series data comprises temperature gradient distribution, gas component concentration and vacuum degree time series fluctuation data in a growth furnace; S2: based on the vacuum control model, a dynamic vacuum threshold curve is generated for each crystal growth stage, and a collaborative control signal is calculated and synchronously output according to the deviation of the real-time acquired current vacuum environment parameter and the dynamic vacuum threshold curve; the collaborative control signal comprises a vacuum pump group start-stop instruction, a micro-leakage compensation instruction and an inert gas injection instruction; S3: according to the collaborative control signal, a three-stage linkage control mode is used to control the vacuum degree fluctuation range in the furnace within a preset range, and a long short-term memory network is used to predict the vacuum degree variation trend; when the prediction deviation exceeds the set vacuum degree variation trend threshold, the standby vacuum unit is activated in advance; S4: in the vacuum degree control process of S3, the distributed optical fiber strain sensor arranged on the furnace body sealing surface collects the sealing surface strain data, uses a leakage acoustic fingerprint identification algorithm to analyze the acoustic characteristics in the sealing surface strain data and locates the failure point, automatically controls the magnetic compensation plug to move to the failure point coordinates to implement dynamic plugging according to the positioning result of the failure point, and feeds back the leakage detection and plugging information to the vacuum control model.

2. The method of claim 1, wherein the vacuum environment is a chamber. The specific steps of S1 comprise: S1.1: real-time acquisition of crystal growth environment time series data in a growth furnace through a sensor arranged in the growth furnace, and pretreatment; the pretreatment comprises cleaning and normalization; S1.2: collection of temperature, vacuum degree and growth stage data in a historical crystal growth process, integration of the pretreated crystal growth environment time series data, establishment of a temperature-vacuum-growth stage three-dimensional mapping relationship through a support vector machine; the growth stage data are obtained by recording the time node and growth stage characteristics of crystal growth; S1.3: experimental measurement of the thermal expansion coefficient of the furnace body material at different temperatures, establishment of a thermal expansion coefficient-temperature relationship database; S1.4: according to the real-time acquired temperature gradient distribution data, the furnace body is divided into N local regions, for each local region, the corresponding thermal expansion coefficient is queried from the thermal expansion coefficient-temperature relationship database according to the temperature value of the local region, the temperature deviation is obtained by calculating the difference between the current temperature and the initial temperature of the local region, and the expansion amount of the local region is obtained by calculating the product of the temperature deviation and the length of the local region and the corresponding thermal expansion coefficient queried; S1.5: the local deformation amount distribution of the furnace body is obtained by calculating the expansion amount of each local region.

3. The method of claim 2, wherein the vacuum environment is a chamber. The specific steps of S1 further comprise: S1.6: According to the structural design drawing of the furnace body, the sealing gap under static working condition is determined, a deformation-sealing gap correction model is constructed based on the expansion amount of each local area and combined with the structural mechanical properties of the furnace body, and then the local deformation distribution of the furnace body is substituted into the deformation-sealing gap correction model to obtain the corrected effective working gap of the vacuum sealing surface; S1.7: According to the corrected effective working gap of the vacuum sealing surface, the dynamic corrected vacuum threshold is obtained in combination with the performance parameters of the vacuum system and the requirements of the crystal growth process; S1.8: The pre-processed crystal growth environment time series data, the temperature-vacuum-growth phase three-dimensional mapping relationship, and the dynamically corrected vacuum threshold are modeled using a multi-layer perception neural network to generate a vacuum control model coupled with multiple parameters.

4. The method of claim 3, wherein the vacuum environment is a chamber. The specific steps of S2 include: S2.1: Obtain the pre-processed crystal growth environment time series data and extract the crystal growth phase characteristic parameters; the crystal growth phase includes the initial stage, the stable growth stage and the end stage; the crystal growth phase characteristic parameters include the temperature change rate, the vacuum degree stability, and the crystal growth time node; S2.2: Input the crystal growth phase characteristic parameters into the vacuum control model, and the vacuum control model outputs the dynamic vacuum threshold curve of the current growth phase; S2.3: Real-time acquisition of the current crystal growth environment time series data through the sensor, calculation of the deviation between the current vacuum degree time series fluctuation data and the dynamic vacuum threshold curve; S2.4: According to the size and direction of the deviation, the start-stop state of the vacuum pump set is determined; If the deviation is greater than the preset deviation threshold, start the vacuum pump set; If the deviation is less than the preset deviation threshold, stop the vacuum pump set; If the deviation is equal to the preset deviation threshold, keep the current vacuum pump set state unchanged; S2.5: According to the size of the deviation, calculate the micro-leak compensation amount; the micro-leak compensation amount is the product of the micro-leak compensation coefficient and the deviation; If the micro-leak compensation amount is greater than zero, increase the micro-leak amount; If the micro-leak compensation amount is less than zero, reduce the micro-leak amount; If the micro-leak compensation amount is equal to zero, maintain the current micro-leak compensation valve state unchanged; S2.6: According to the deviation and the current gas component concentration, calculate the inert gas injection amount; the inert gas injection amount is the difference between the target gas component concentration and the current gas component concentration, multiplied by the inert gas injection coefficient and the deviation; If the inert gas injection amount is greater than zero, inject inert gas; If the inert gas injection amount is less than zero, reduce the inert gas injection amount; If the inert gas injection amount is equal to zero, keep the inert gas injection valve in the current state; S2.7: Integrate the vacuum pump set start-stop instruction, the micro-leak compensation instruction, and the inert gas injection instruction, and output the integrated collaborative control signal to the actuator, including the vacuum pump set controller, the micro-leak compensation valve, and the inert gas injection valve.

5. The method of claim 4, wherein the method further comprises: The three-stage linkage control mode specifically includes: The first stage control is the coarse control of the vacuum pump set, which controls the opening and closing number of the vacuum pump set according to the vacuum pump set start-stop instruction in the collaborative control signal to adjust the vacuum degree in the furnace; The second level control is a fine adjustment control of the micro-leakage compensation valve, and the flow of the micro-leakage compensation valve is adjusted according to the micro-leakage compensation instruction to finely adjust the vacuum degree in the furnace; The third level control is an auxiliary control of the inert gas injection valve, and the injection amount and injection speed of the inert gas are controlled according to the inert gas injection instruction to stabilize the vacuum degree in the furnace.

6. The method of claim 5, wherein the vacuum environment is a chamber. The advance activation of the standby vacuum unit when the prediction deviation exceeds the set vacuum degree change trend threshold value comprises: The vacuum degree change trend threshold value is set according to the crystal growth process requirements and the equipment performance; The vacuum degree change trend predicted by the long short-term memory network is compared with the set vacuum degree change trend threshold value; If the prediction deviation is greater than the vacuum degree change trend threshold value, an activation instruction is sent to the standby vacuum unit to start the standby vacuum unit in advance to increase the pumping capacity of the vacuum system; If the prediction deviation is less than the opposite of the vacuum degree change trend threshold value, an activation instruction is sent to the standby vacuum unit to start the standby vacuum unit in advance to reduce the pumping capacity of the vacuum system or increase the inert gas injection; If the prediction deviation is equal to the vacuum degree change trend threshold value, the current control state is kept unchanged.

7. The method of claim 6, wherein the method further comprises: The specific steps of the S4 comprise: S4.1: M distributed optical fiber strain sensors are arranged around the furnace body sealing surface, the spatial coordinates of each sensor are determined, and a sealing surface coordinate system is established; S4.2: A control system of the magnetic compensation plug is established, and at the same time, a mapping relationship between the sealing surface coordinate system and the mechanical arm motion coordinate is established; the control system of the magnetic compensation plug comprises a driving motor, a transmission mechanism and a position sensor; S4.3: When the furnace body sealing surface is in a normal working condition, the distributed optical fiber strain sensors are used to continuously collect the sealing surface strain data within ΔT time, and the collected data is preprocessed to obtain preprocessed normal working condition data; the normal working condition is no leakage; S4.4: The Fourier transform is used to extract the acoustic wave characteristics of the preprocessed normal working condition data, and the extracted acoustic wave characteristics are stored in the sealing surface normal working condition acoustic wave characteristic library; S4.5: The distributed optical fiber strain sensors arranged on the furnace body sealing surface are used to collect the strain data of the sealing surface in real time, and the collected data is preprocessed to obtain preprocessed strain data; S4.6: The preprocessed strain data is analyzed by the wavelet packet transform to calculate the energy of each frequency band, and the calculated energy is compared with the frequency band energy in the sealing surface normal working condition acoustic wave characteristic library; When the energy of any frequency band deviates from the normal range, it is determined that the frequency band contains an abnormal acoustic wave signal, and the leakage positioning process is entered; the abnormal acoustic wave signal refers to an acoustic wave leakage signal.

8. The method of claim 7, wherein the method further comprises: The specific steps of the S4 further comprise: S4.7: For the extracted abnormal acoustic wave signal, the cross-correlation function between the signal received by different sensors is calculated respectively, and the time difference of the arrival of the acoustic wave at different sensors is determined by finding the peak position of the cross-correlation function; S4.8: According to the time difference of the arrival of the acoustic wave at different sensors, combined with the spatial coordinates of each sensor, a positioning equation set is constructed, and the three-dimensional coordinates of the failure point are calculated by solving the positioning equation set. S4.9: Convert the three-dimensional coordinates of the failure point into control coordinates in the mechanical arm motion coordinate system using the mapping relationship between the sealing surface coordinate system and the mechanical arm motion coordinate; S4.10: Based on the control coordinates in the mechanical arm motion coordinate system, send control instructions to the drive motor to control the drive motor to rotate according to the predetermined motion trajectory and speed, drive the magnetic compensation plug to move through the transmission mechanism, and plan the movement path during the movement of the magnetic compensation plug using the path planning algorithm; S4.11: Real-time feedback of the position information of the magnetic compensation plug through the position sensor, and comparison with the target position in the control instruction, the control system adjusts the motion of the drive motor in real time according to the comparison result.

9. The method of claim 8, wherein the method further comprises: The specific steps of S4 also include: S4.12: When the magnetic compensation plug reaches the three-dimensional coordinates of the failure point, the control system controls the magnetic compensation plug to generate a magnetic field, and makes the magnetic compensation plug tightly fit with the furnace body sealing surface under the action of the magnetic field, fills the leakage point of the sealing surface, and completes the dynamic plugging of the leakage point; S4.13: Verify the plugging effect by monitoring the change of the vacuum degree in the furnace; If the vacuum degree is stable within the normal range, it means that the plugging is successful; If the vacuum degree is abnormal, adjust the position or magnetic field strength of the magnetic compensation plug; S4.14: Feedback the leakage detection and plugging information to the vacuum control model to make the vacuum control model adjust the control strategy according to the actual situation and optimize the vacuum degree control effect; The leakage detection and plugging information includes leakage type, failure point position, and plugging effect.

10. A linkage regulation system of vacuum environment, used for implementing the linkage regulation method of vacuum environment in any one of claims 1-9, characterized in that, Including: Data acquisition module, cooperative control module, trend prediction module, leakage detection module; The data acquisition module is used to collect the crystal growth environment data in real time, and to build a multi-parameter coupled vacuum control model; The cooperative control module generates a dynamic vacuum threshold curve based on the vacuum control model and calculates a cooperative control signal; The trend prediction module maintains the stability of the vacuum degree through a three-stage linkage control method and predicts the trend of the vacuum degree change; The leakage detection module is used to detect the leakage of the furnace body sealing surface and locate the failure point, and to implement dynamic plugging.

Images