Material growth parameter adjustment system and method
By designing data acquisition, processing and storage modules in the material growth system, combined with the FPGA's extensible model interface, the problem of low model storage and processing efficiency is solved, the rapid and accurate control of material growth parameters is achieved, and the material growth is improved Quality and yield.
Patent Information
- Application Number
- CN202410254224.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-03-06
- Publication Date
- 2025-01-03
- Estimated Expiration
- 2044-03-06
AI Technical Summary
In the process of material growth, the storage and processing of models are inefficient and prone to memory overflow and lag, especially in large-scale model processing and real-time data processing, and there is a lack of effective solutions.
设计了一种材料生长参数的调整系统，包括数据获取模块、处理模块、模型调用模块、模型数据再处理模块和参数调整模块，通过可扩充模型接口与现场可编程逻辑门阵列FPGA连接，实现快速处理和实时存储，降低数据量，提高处理效率。
It improves the real-time control and timely data processing efficiency of material growth parameters, improves the quality of material growth and yield rate, avoids system lag, and ensures the efficiency and accuracy of data processing.
Smart Images
Figure CN118609713B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of material preparation, and in particular to a system and method for adjusting material growth parameters. Background Art
[0002] The existing hardware architectures for machine learning deployment are mainly CPU, GPU, TPU and FPGA architectures. CPU is generally applicable to small models and limits high power consumption applications. GPU and TPU are applicable to large-scale matrix operations and tensor reasoning. FPGA is customized and optimized for specific tasks. In the computing architecture, CPU, GPU and TPU need to store the model in the file system. The update and call of the model depend on computer storage. Especially for large models, the transmission and mirroring of such model files to the local computer takes a long time. For FPGA, although it embeds the model into the hardware architecture and does not require additional model storage, it is limited by the programmable content of the hardware and can only be applied to specific tasks, lacking versatility. In addition, the computer running speed needs to be taken into account. During the operation of the computer, it needs to monitor the changes in material growth parameters and real-time control, obtain the image or curve data of the in-situ characterization device in real time, call the model to accelerate the rapid processing of the model data, and all the acquired data needs to be stored in the hard disk in real time, which is heavy for the hard disk and computing processing load, making the computer prone to memory overflow errors. In model processing, a large model is repeatedly called and processed, which will make the information bandwidth on a single storage device easy to occupy, causing the computer to freeze and fail. Summary of the invention
[0003] This application proposes a system and method for adjusting material growth parameters. The specific scheme is as follows:
[0004] In one aspect, an embodiment of the present application provides a material growth parameter adjustment system, including: a data acquisition module, a processing module, a model calling module, a model data reprocessing module, a parameter adjustment module, and an expandable model interface;
[0005] Wherein, the data acquisition module is used to acquire a first parameter value corresponding to a growth parameter in a material growth device and in-situ characterization data corresponding to a growing material collected by an in-situ characterization device;
[0006] The processing module is used to determine a target parameter acquisition model from a plurality of parameter acquisition models according to a target type and a target growth stage of the growing material, wherein the parameter acquisition model is connected to the adjustment system via the expandable model interface;
[0007] The model calling module is used to input at least one of the first parameter value and the in-situ characterization data into the target parameter acquisition model through the expandable model interface corresponding to the target parameter acquisition model, and obtain an output result of the target parameter acquisition model;
[0008] The model data reprocessing module is used to determine a second parameter value corresponding to the growth parameter according to the output result;
[0009] The parameter adjustment module is used to adjust the parameter value corresponding to the growth parameter based on the second parameter value.
[0010] In some embodiments, the processing module is used to:
[0011] Obtaining the growth time corresponding to the growing material;
[0012] Based on the growth time, querying a growth stage mapping table associated with the target type to obtain a target growth stage of the growing material;
[0013] The target type and the target growth stage are matched with the identifier corresponding to each of the parameter acquisition models to determine the target parameter acquisition model from the multiple parameter acquisition models, wherein the identifier corresponding to the parameter acquisition model includes the type and growth stage of the material.
[0014] In some embodiments, the model data reprocessing module is used to:
[0015] Obtaining a growth target corresponding to the growing material at a current moment;
[0016] Determining an adjustment direction corresponding to the growth parameter according to a difference between the output result and the growth target;
[0017] Based on the adjustment direction corresponding to the growth parameter and the preset parameter adjustment value, a second parameter value corresponding to the growth parameter is determined.
[0018] In some embodiments, the system further comprises a data feature preprocessing module, wherein the data feature preprocessing module is connected to the data acquisition module;
[0019] The data feature preprocessing module is used to preprocess the parameter value corresponding to the growth parameter and / or the in-situ characterization data.
[0020] In some embodiments, the system further comprises a real-time data storage module, wherein the real-time data storage module is used to store parameter values corresponding to the growth parameters, the in-situ characterization data, and the output results.
[0021] In some embodiments, the parameter acquisition model is deployed in a field programmable gate array (FPGA).
[0022] In some embodiments, the type of the parameter acquisition model is a prediction model or a discriminant model.
[0023] In some embodiments, the model data reprocessing module is used to:
[0024] In the case where there are multiple target parameter acquisition models, fusing the multiple output results to obtain the target result;
[0025] A second parameter value corresponding to the growth parameter is determined according to the target result.
[0026] Another aspect of the present application provides a method for adjusting material growth parameters, including:
[0027] Acquire a first parameter value corresponding to a growth parameter in a material growth device and in-situ characterization data corresponding to a growing material collected by an in-situ characterization device;
[0028] Determining a target parameter acquisition model from a plurality of parameter acquisition models according to a target type and a target growth stage of the growing material, wherein the parameter acquisition model is connected to the adjustment system via the expandable model interface;
[0029] Inputting at least one of the first parameter value and the in-situ characterization data into the target parameter acquisition model through an expandable model interface corresponding to the target parameter acquisition model, and acquiring an output result of the target parameter acquisition model;
[0030] Determining a second parameter value corresponding to the growth parameter according to the output result;
[0031] Based on the second parameter value, the parameter value corresponding to the growth parameter is adjusted.
[0032] In some embodiments, determining a target parameter acquisition model from a plurality of parameter acquisition models according to a target type and a target growth stage of the growing material includes:
[0033] Obtaining the growth time corresponding to the growing material;
[0034] Based on the growth time, querying a growth stage mapping table associated with the target type to obtain a target growth stage of the growing material;
[0035] The target type and the target growth stage are matched with the identifier corresponding to each of the parameter acquisition models to determine the target parameter acquisition model from the multiple parameter acquisition models, wherein the identifier corresponding to the parameter acquisition model includes the type and growth stage of the material.
[0036] In the disclosed embodiment, the parameter acquisition model is connected to the adjustment system in the form of a plug-in through an expandable model interface, so that when the adjustment system determines the target parameter acquisition module corresponding to the growing material, it can directly call the external target parameter acquisition model, process the first parameter value and in-situ characterization data corresponding to the growth parameter, and obtain the output result of the target parameter acquisition model, and then adjust the parameter value corresponding to the growth parameter according to the output result. Therefore, the amount of data processed by the adjustment system can be reduced by means of an external parameter acquisition model, avoiding system freezes, thereby improving data processing efficiency, thereby improving the timeliness of growth parameter control, and improving material growth quality and yield rate.
[0037] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS
[0038] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:
[0039] Figure 1 A schematic diagram of the structure of a material growth parameter adjustment system provided in an embodiment of the present application;
[0040] Figure 2 A schematic diagram of the structure of a material growth parameter adjustment system provided in an embodiment of the present application;
[0041] Figure 3 A schematic flow chart of another method for adjusting material growth parameters provided in an embodiment of the present application. DETAILED DESCRIPTION
[0042] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.
[0043] The material growth parameter adjustment method of the embodiment of the present application is performed by the material growth parameter adjustment device provided by the embodiment of the present application (hereinafter referred to as the adjustment device), which can be configured in a computer device to improve the material growth quality and yield rate.
[0044] The following describes a system for adjusting material growth parameters according to an embodiment of the present application with reference to the accompanying drawings.
[0045] Figure 1 A schematic diagram of the structure of a material growth parameter adjustment system provided in an embodiment of the present application.
[0046] like Figure 1 As shown, the material growth parameter adjustment system includes: a data acquisition module 101, a processing module 102, a model calling module 103, a model data reprocessing module 104, a parameter adjustment module 105, and an expandable model interface 106;
[0047] The data acquisition module 101 is used to acquire a first parameter value corresponding to a growth parameter in a material growth device and in-situ characterization data corresponding to a growing material collected by an in-situ characterization device;
[0048] A processing module 102, for determining a target parameter acquisition model from a plurality of parameter acquisition models according to a target type and a target growth stage of the growing material, wherein the parameter acquisition model is connected to the adjustment system via an expandable model interface 106;
[0049] The model calling module 103 is used to input at least one of the first parameter value and the in-situ characterization data into the target parameter acquisition model through an expandable model interface corresponding to the target parameter acquisition model, and obtain an output result of the target parameter acquisition model;
[0050] A model data reprocessing module 104, used to determine a second parameter value corresponding to the growth parameter according to the output result;
[0051] The parameter adjustment module 105 is used to adjust the parameter value corresponding to the growth parameter based on the second parameter value.
[0052] The material growth equipment may be a semiconductor material growth equipment, and the semiconductor material growth equipment may be a molecular beam epitaxy (MBE) or a metal-organic chemical vapor deposition (Metal-Organic Chemical Vapor Deposition).
[0053] The growth parameters may be all parameters in the material growth equipment that affect the material growth, such as shutter, temperature, vacuum, etc. This disclosure does not limit this.
[0054] The in-situ characterization equipment may be one or more of a reflection high-energy electron diffraction (RHEED), a quadrupole mass spectrometer (QMS), etc. The present disclosure does not limit this.
[0055] The in-situ characterization data may be data obtained by an electron beam acting on a material surface, for example, optical spectrum data, mass spectrum data, or a combination thereof. The in-situ characterization data may be in the form of one or more of picture data, video data, and curve data.
[0056] It should be noted that a device may have a multi-layer structure, and the materials of each layer are different. Among them, the growing material may be a layer of material in the device being prepared, and the target type of the material may be indium arsenide (InAs), indium phosphide (InP), gallium arsenide (GaAs), gallium nitride (GaN), aluminum gallium arsenide (AlGaAs), etc.
[0057] In some embodiments, the target type of the growing material can be determined based on the shutter state of the growth source, or the shutter state of the growth source and the growth time, wherein the shutter state refers to whether the switch of the growth source is on or off.
[0058] In some embodiments, each layer of material can be divided into multiple growth stages, for example, it can be divided into an initial growth stage, a growth process stage, a growth completion stage, etc. The stage in which the growing material is currently located is the target growth stage.
[0059] In some embodiments, each growth stage of each material may correspond to one or more parameter acquisition models. The multiple parameter acquisition models may be models for adjusting the same growth parameters but with different model parameters. The parameter acquisition models may be generated in advance based on sample data of each material at each stage.
[0060] In some embodiments, the parameter acquisition model is deployed in a field programmable gate array (FPGA). Optionally, one parameter acquisition model or multiple parameter acquisition models may be deployed in the FPGA, which is not limited in the present disclosure.
[0061] In some embodiments, the FPGA is externally connected to the adjustment system through the expandable model interface 106. In some embodiments, the FPGA is pluggable. Thus, the adjustment system 100 can call the parameter acquisition model deployed in the FPGA through the expandable model interface 106. And by using modular and expandable model hardware plug-ins, it is possible to customize the corresponding model module for a special material system without the need to modify the input and output of the model in large quantities to adapt to special material growth equipment.
[0062] In some embodiments, the FPGA in which the parameter acquisition model is deployed can be regarded as a plug-in. This plug-in is similar to an external FPGA, and has a machine learning model inside. A common protocol is defined between the API and the adjustment system 100. The adjustment system 100 can pass the data required by the FPGA to this plug-in through the expandable model interface 106, and read the output result of the plug-in. This type of FPGA plug-in can be combined with Docker to encapsulate the model, model dependencies, and data processing environment in an independent container to ensure consistency and portability in different environments without being affected by the adjustment system.
[0063] This design of FPGA as a pluggable hardware module utilizes the reconfigurability of FPGA, dynamically configures the logic and computing units in FPGA according to the needs of the specific material system, and connects to the computer through PCI Express or other high-speed interfaces. Such a module can provide dedicated computing resources without occupying too much CPU and memory resources of the adjustment system 100, thereby giving full play to the advantages of hardware acceleration.
[0064] In some embodiments, the type of the parameter acquisition model is a prediction model or a discrimination model. It should be noted that different growth stages of each material may correspond to different types of parameter acquisition models.
[0065] Among them, the prediction model is a growth result-oriented model. For example, in the process of quantum dot growth, when the InAs quantum dots reach a density of 4-6E10cm-2 and the best uniformity, the optimal luminous intensity and wavelength correspond to the best. The formation of quantum dots is the Stranski-Krastanov (SK) growth mode. Quantum dots are formed in the stress release stage, and the accumulation of stress in the stress accumulation stage will affect the result of stress release. This surface quantum dot arrangement can be observed by in-situ RHEED. By analyzing the possibility of predicting the stress release result in the stress accumulation stage, the model gives the predicted distribution results of quantum dots after stress release, as well as the substrate temperature and As needle valve size that need to be adjusted, to optimize the stress accumulation process, so that the probability of outputting the best quantum dot surface arrangement gradually increases.
[0066] Among them, the discriminant model is a model guided by the current growth state. It gives the parameter adjustment results based on the current growth state. Continuing with the above-mentioned InAs quantum dot SK growth mode, quantum dots will gradually form in the stress release stage, and will experience the following stages: no quantum dots, quantum dot density gradually increases, quantum dot distribution reaches the best, quantum dot density reaches the best, quantum dot distribution reaches the second best, quantum dot density gradually decreases, quantum dot density is stable and quantum dot size increases. Obviously, the end of the quantum dot growth state should be completed when the quantum dot distribution reaches the best and the quantum dot density reaches the best. The shutter is closed immediately. If the baffle cannot be closed in time, the quality of the quantum dots will deteriorate. Therefore, the model needs to accurately identify which stage the quantum dot growth is in, and output the shutter control state in time to close the shutter in time.
[0067] The above is based on the example of InAs quantum dots, whose growth process requires two models to be implemented. For thick GaAs materials, if 4×4 reconstruction lines are observed through RHEED or the 2×4 reconstruction lines become more blurred, it means that the surface roughness of the material increases and the quality of material growth decreases. Therefore, only a single model that adjusts parameters based on the judgment results (i.e., a model of discriminating type) is required. The model adjusts the size of the As needle valve by judging the surface reconstruction state of the material growth. If the state of the surface reconstruction line 2×4 is too blurred or there is a 4×4 reconstruction line, it is necessary to immediately increase the As needle valve so that the material always grows in a 2×4 reconstruction state.
[0068] In some embodiments, each growth stage of each material may correspond to multiple parameter acquisition models, and the output result of each parameter acquisition model is used to adjust different growth parameters.
[0069] For example, taking GaAs as an example, the above problem only explains the model design from the perspective of thick GaAs film. GaAs materials can also detect growth thickness and growth rate. For AlGaAs materials, the component distribution of each element and each layer with different thicknesses inside the material also needs to be considered. The GaAs growth rate is optimal at 0.6-0.7ML / s. Through in-situ RHEED interval monitoring, the model determines the current growth rate and the beam intensity corresponding to the Ga source furnace temperature that needs to be optimized. For AlGaAs or InGaAs materials, AlGaAs needs to be grown to achieve the best luminescence effect. 0.3 Ga 0.7 As and In 0.15 Ga 0.85 For fixed thickness applications, when growing Bragg reflector-type surface-emitting lasers, the thickness of the multi-layer GaAs and AlGaAs cross-structures needs to be strictly controlled to be equal to ensure that the laser resonates best in the cavity when the device is used.
[0070] In some embodiments, based on the strictness of the requirements for growth parameters, the discrimination type can be further divided into long-time discrimination type and short-time discrimination type. Taking GaAs reconstruction transformation as an example, even if 4×4 reconstruction occurs, by adding an As needle valve within 1 minute, the material surface can still be restored to a growth state of only 2×4. This control process can span a period of 1 minute or even longer, which belongs to long-time discrimination. Time series models can be used for application, such as long short-term memory neural networks, recurrent neural networks, etc. For GaN materials, the RHEED spot shows point-like characteristics, the material has been severely degraded, and a smooth film cannot be grown. Even if a nitrogen (N) gas phase source is added, it cannot be restored, which belongs to short-time discrimination. Therefore, during the growth of GaN materials, if a RHEED spot appears faintly or the RHEED stripes are unclear, it is necessary to adjust the flow rate of the N-containing gas phase source in time to curb the probability of degradation development.
[0071] In some embodiments, the processing module 102 is used to: obtain the growth time corresponding to the growing material; based on the growth time, query the growth stage mapping table associated with the target type to obtain the target growth stage of the growing material; match the target type and target growth stage with the identifier corresponding to each parameter acquisition model to determine the target parameter acquisition model from multiple parameter acquisition models, wherein the identifier corresponding to the parameter acquisition model includes the type and growth stage of the material.
[0072] In the disclosed embodiment, the growth stage corresponding to each type of material at each growth moment can be determined based on empirical design, and then a growth stage mapping table can be constructed based on the growth stage corresponding to each type of material at each growth moment. Thus, the pre-set growth stage mapping table can be queried based on the growth time corresponding to the growing material to determine the target growth stage.
[0073] When the FPGA where each parameter acquisition model is located is connected to the adjustment system 100, the identifier of the newly connected parameter acquisition model can be read and stored, so that when obtaining the target growth stage of the growing material, it can be quickly queried whether there is a suitable target parameter acquisition model.
[0074] In some embodiments, the model data reprocessing module is used to: obtain the growth target corresponding to the growing material at the current moment, then determine the adjustment direction corresponding to the growth parameter based on the difference between the output result and the growth target, and finally determine the second parameter value corresponding to the growth parameter based on the adjustment direction corresponding to the growth parameter and the preset parameter adjustment value.
[0075] The adjustment direction includes increasing and decreasing. The preset parameter adjustment value may be an adjustment value corresponding to a preset growth parameter. For example, the parameter adjustment value corresponding to the temperature may be 1 degree Celsius (°C), 2 degrees Celsius (°C), etc. This disclosure does not limit this.
[0076] In some embodiments, the model data reprocessing module is used to: when there are multiple target parameter acquisition models, fuse multiple output results to obtain the target result; and determine the second parameter value corresponding to the growth parameter according to the target result.
[0077] Among them, the multiple target parameter acquisition models are models for adjusting the same growth parameters. Therefore, the output results of the multiple target parameter acquisition models can be fused to reduce errors, improve the accuracy of the acquired target results, and increase the adjustment efficiency.
[0078] Specifically, the multiple output results may be fused by taking the average of the multiple output results. Alternatively, different weights may be assigned to each target parameter acquisition model, and then the weighted sum of the multiple output results is taken as the target result. The present disclosure does not specifically limit the method for fusing multiple output results.
[0079] In the disclosed embodiment, the parameter acquisition model is connected to the adjustment system in the form of a plug-in through an expandable model interface, so that when the adjustment system determines the target parameter acquisition module corresponding to the growing material, it can directly call the external target parameter acquisition model, process the first parameter value and in-situ characterization data corresponding to the growth parameter, and obtain the output result of the target parameter acquisition model, and then adjust the parameter value corresponding to the growth parameter according to the output result. Therefore, the amount of data processed by the adjustment system can be reduced by means of an external parameter acquisition model, avoiding system freezes, thereby improving data processing efficiency, thereby improving the timeliness of growth parameter control, and improving material growth quality and yield rate.
[0080] Figure 2 A schematic diagram of the structure of a material growth parameter adjustment system provided in an embodiment of the present application.
[0081] like Figure 2 As shown, the material growth parameter adjustment system also includes a data feature preprocessing module 107, which is connected to the data acquisition module 101; the data feature preprocessing module 107 is used to preprocess the parameter values corresponding to the growth parameters and / or the in-situ characterization data.
[0082] Among them, the feature preprocessing module is set in the adjustment system 100, so that the parameter values and / or in-situ characterization data corresponding to the growth parameters can be preprocessed before the parameter values and / or in-situ characterization data corresponding to the growth parameters are input into the target parameter acquisition model, so that there is no need to preprocess the data in the model, thereby improving the processing efficiency of the data acquisition model.
[0083] like Figure 2 As shown, the material growth parameter adjustment system also includes a real-time data storage module 108, which is used to store parameter values corresponding to the growth parameters, in-situ characterization data, and output results. Therefore, the parameter values corresponding to the growth parameters, in-situ characterization data, and output results can be stored through the real-time data storage module 108, thereby not only avoiding data loss, but also querying historical data.
[0084] Figure 3 The present invention provides a flow chart of a method for adjusting material growth parameters, which is performed by a material growth parameter adjustment system. Figure 3 As shown, the method for adjusting the material growth parameters includes:
[0085] Step 301, obtaining a first parameter value corresponding to a growth parameter in a material growth device and in-situ characterization data corresponding to a growing material collected by an in-situ characterization device.
[0086] Step 302, determining a target parameter acquisition model from a plurality of parameter acquisition models according to a target type and a target growth stage of the growing material, wherein the parameter acquisition model is connected to the adjustment system via an expandable model interface.
[0087] In some embodiments, according to the target type and target growth stage of the growing material, a target parameter acquisition model is determined from a plurality of parameter acquisition models, including:
[0088] Get the growth time corresponding to the growing material;
[0089] Based on the growth time, query the growth stage mapping table associated with the target type to obtain the target growth stage of the growing material;
[0090] The target type and target growth stage are matched with the identifier corresponding to each parameter acquisition model to determine the target parameter acquisition model from multiple parameter acquisition models, wherein the identifier corresponding to the parameter acquisition model includes the type and growth stage of the material.
[0091] Step 303: input at least one of the first parameter value and the in-situ characterization data into the target parameter acquisition model through an expandable model interface corresponding to the target parameter acquisition model, and obtain an output result of the target parameter acquisition model.
[0092] Step 304: Determine a second parameter value corresponding to the growth parameter according to the output result.
[0093] Step 305: Based on the second parameter value, adjust the parameter value corresponding to the growth parameter.
[0094] It should be noted that the above explanations and effects of the embodiment of the system for adjusting material growth parameters are also applicable to the method for adjusting material growth parameters of this embodiment, so they will not be repeated here.
[0095] In order to implement the above embodiment, the embodiment of the present application also proposes a computer device, including a processor and a memory;
[0096] The processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to implement the material growth parameter adjustment method as described in the above embodiment.
[0097] In order to implement the above embodiments, the embodiments of the present application further provide a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the method for adjusting the material growth parameters as described in the above embodiments is implemented.
[0098] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in the field can change, modify, replace and modify the above embodiments within the scope of the present application.
Claims
1. A material growth parameter adjustment system, characterized in that: include: Data acquisition module, processing module, model calling module, model data reprocessing module, parameter adjustment module, expandable model interface; Wherein, the data acquisition module is used to acquire a first parameter value corresponding to a growth parameter in a material growth device and in-situ characterization data corresponding to a growing material collected by an in-situ characterization device; The processing module is used to obtain the growth time corresponding to the growing material, query the growth stage mapping table associated with the target type of the growing material based on the growth time to obtain the target growth stage of the growing material, match the target type and the target growth stage with the identifier corresponding to each parameter acquisition model to determine the target parameter acquisition model from the multiple parameter acquisition models, wherein the identifier corresponding to the parameter acquisition model includes the type and growth stage of the material, and the parameter acquisition model is connected to the adjustment system through the expandable model interface; The model calling module is used to input at least one of the first parameter value and the in-situ characterization data into the target parameter acquisition model through the expandable model interface corresponding to the target parameter acquisition model, and obtain an output result of the target parameter acquisition model; The model data reprocessing module is used to determine a second parameter value corresponding to the growth parameter according to the output result; The parameter adjustment module is used to adjust the parameter value corresponding to the growth parameter based on the second parameter value.
2. The system according to claim 1, characterized in that The model data reprocessing module is used to: Obtaining a growth target corresponding to the growing material at a current moment; Determining an adjustment direction corresponding to the growth parameter according to a difference between the output result and the growth target; Based on the adjustment direction corresponding to the growth parameter and the preset parameter adjustment value, a second parameter value corresponding to the growth parameter is determined.
3. The system according to any one of claims 1-2, characterized in that: The system further comprises a data feature preprocessing module, wherein the data feature preprocessing module is connected to the data acquisition module; The data feature preprocessing module is used to preprocess the parameter value corresponding to the growth parameter and / or the in-situ characterization data.
4. The system according to any one of claims 1-2, characterized in that: The system also includes a real-time data storage module, which is used to store parameter values corresponding to the growth parameters, the in-situ characterization data, and the output results.
5. The system according to any one of claims 1-2, characterized in that: The parameter acquisition model is deployed in a field programmable gate array FPGA.
6. The system according to any one of claims 1-2, characterized in that: The type of the parameter acquisition model is a prediction model or a discriminant model.
7. The system according to any one of claims 1-2, characterized in that: The model data reprocessing module is used to: In the case where there are multiple target parameter acquisition models, fusing the multiple output results to obtain the target result; A second parameter value corresponding to the growth parameter is determined according to the target result.
8. A method for adjusting material growth parameters, performed by the material growth parameter adjustment system according to any one of claims 1 to 7, characterized in that: The method comprises: Acquire a first parameter value corresponding to a growth parameter in a material growth device and in-situ characterization data corresponding to a growing material collected by an in-situ characterization device; Obtaining the growth time corresponding to the growing material; Based on the growth time, querying a growth stage mapping table associated with the target type of the growing material to obtain a target growth stage of the growing material; Matching the target type and the target growth stage with an identifier corresponding to each parameter acquisition model to determine a target parameter acquisition model from a plurality of the parameter acquisition models, wherein the identifier corresponding to the parameter acquisition model includes a type and a growth stage of a material, and the parameter acquisition model is connected to the adjustment system via an expandable model interface; Inputting at least one of the first parameter value and the in-situ characterization data into the target parameter acquisition model through an expandable model interface corresponding to the target parameter acquisition model, and acquiring an output result of the target parameter acquisition model; Determining a second parameter value corresponding to the growth parameter according to the output result; Based on the second parameter value, the parameter value corresponding to the growth parameter is adjusted.