Crystal pulling methods, systems, computer equipment, and storage media based on big data analytics

Abstract

This invention relates to a crystal pulling method, system, computer equipment, and storage medium based on big data analysis. The steps are as follows: S1: Acquire basic source data for multiple nodes of each workpiece in each single crystal furnace during the crystal pulling process; S2: Process the acquired source data, filtering and converting it into several easily identifiable and labelable parameters for each node of each workpiece, and obtaining a dataset of all parameter values ​​for each node of the workpiece; S3: Build a model for each parameter in each node of each workpiece using deep learning; S4: Compare the values ​​of each parameter in each node of each workpiece in S2 with the model in S3 to determine whether the values ​​of each parameter in the node of the workpiece are reasonable. This invention can effectively perform intelligent crystal pulling using big data and deep learning, utilizing big data analysis and executing optimization schemes, and then organically combining big data and deep learning to improve crystal pulling quality and efficiency while reducing crystal pulling costs.

Description

Technical Field

[0001] This invention belongs to the field of photovoltaic monocrystalline silicon pulling technology, and in particular relates to crystal pulling methods, systems, computer equipment and storage media based on big data analysis. Background Technology

[0002] The Czochralski single crystal growth process mainly includes temperature stabilization, crystal pulling, shoulder formation, diameter equalization, and finishing, among other steps. Currently, the single crystal pulling process is mainly controlled by a manually operated centralized control system. Although this control method can achieve automated operation within a standard range through the centralized control system, the decision-making part still requires professional engineers to operate and control. This crystal pulling method results in unstable production, low crystal pulling efficiency, and too many human factors, leading to high crystal pulling production costs. Summary of the Invention

[0003] This invention provides a crystal pulling method, system, computer equipment, and storage medium based on big data analysis, which is particularly suitable for solar Czochralski silicon single crystal production, and solves the technical problems of unstable crystal pulling quality and high production cost caused by manual operation and centralized control system in the prior art.

[0004] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0005] The crystal pulling method based on big data analysis includes the following steps:

[0006] S1: Obtain the basic source data of multiple nodes for each workpiece in each single crystal furnace during the crystal pulling process;

[0007] S2: Process the acquired source data, filter and convert it into several parameters that are easy to identify and label in each node of each workpiece, and obtain a dataset of all parameter values ​​for each node of the workpiece;

[0008] S3: Build a model for each parameter in each node of each workpiece using deep learning;

[0009] S4: Compare the value of each parameter in each node of each workpiece in S2 with the model in S3 to determine whether the value of each parameter in the node where the workpiece is located is reasonable.

[0010] Furthermore, in S4, if all the parameter values ​​in the node where the workpiece is located are within the range of the model, then the operation of the next node continues;

[0011] If some of the parameter values ​​in the node where the workpiece is located are not within the range of the model, then pause or adjust the relevant parameters in the corresponding node.

[0012] Furthermore, all parameter types of each workpiece and each node in S2 correspond to all parameter types of each workpiece and each node in S3;

[0013] Preferably, the parameters are established according to the production area, the start and end times of each node, the associated data between upstream and downstream nodes, and the different functions in each node;

[0014] Preferably, all the parameters of each node in each workpiece are configured and displayed on the terminal display of the single crystal furnace in which the workpiece is drawn.

[0015] Furthermore, the source data for each node in each workpiece includes production process data and / or raw material data and / or quality data.

[0016] Preferably, the nodes in each workpiece include at least a temperature stabilization node, a crystal-leading node, a shoulder-forming node, a constant-diameter node, and a tailing node, and the temperature stabilization node, the crystal-leading node, the shoulder-forming node, the constant-diameter node, and the tailing node are arranged sequentially.

[0017] Preferably, the model includes at least a steady-temperature node model, a crystal-leading node model, a shoulder-forming node model, a constant-diameter node model, and a tail-end node model.

[0018] A crystal pulling system, the system comprising:

[0019] Source data acquisition unit: used to acquire the basic source data of multiple nodes of each workpiece in each single crystal furnace during the crystal pulling process;

[0020] Source data processing unit: used to process the acquired source data, filter and convert it into several parameters that are easy to identify and label in each node of each workpiece, and obtain a dataset of all parameter values ​​of each node of the workpiece;

[0021] Model building unit: used to build a model for each parameter in each node of each workpiece through deep learning;

[0022] Parameter determination unit: used to compare the value of each parameter in each node of each workpiece in the processing source data unit with the model in the model building unit to determine whether the value of each parameter in the node where the workpiece is located is reasonable.

[0023] Furthermore, in the determination parameter unit, if all the parameter values ​​in the node where the workpiece is located are within the range of the model, then the operation of the next node continues;

[0024] If some of the parameter values ​​in the node where the workpiece is located are not within the range of the model, then the process will pause or adjust the relevant parameters in the corresponding node.

[0025] Furthermore, all parameter types of each workpiece and each node in the processing source data unit correspond to all parameter types of each workpiece and each node in the model building unit;

[0026] Preferably, the parameters are established according to the production area, the start and end times of each node, the associated data between upstream and downstream nodes, and the different functions in each node;

[0027] Preferably, all the parameters of each node in each workpiece are configured and displayed on the terminal display of the single crystal furnace in which the workpiece is drawn.

[0028] Furthermore, the source data for each node in each workpiece includes production process data and / or raw material data and / or quality data.

[0029] Preferably, the nodes in each workpiece include at least a temperature stabilization node, a crystal-leading node, a shoulder-forming node, a constant-diameter node, and a tailing node, and the temperature stabilization node, the crystal-leading node, the shoulder-forming node, the constant-diameter node, and the tailing node are arranged sequentially.

[0030] Preferably, the model includes at least a steady-temperature node model, a crystal-leading node model, a shoulder-forming node model, a constant-diameter node model, and a tail-end node model.

[0031] A computer device includes a memory and a processor; the memory stores a computer program; the processor is configured to execute the computer program, and when executing the computer program, cause the processor to perform the steps of the crystal pulling method as described in any of the preceding claims.

[0032] A computer-readable storage medium storing a computer program that, when executed by the processor, causes the processor to perform the steps of the crystal pulling method as described in any of the preceding claims.

[0033] Compared with existing technologies, the crystal pulling method, system, computer equipment, and storage medium based on big data analysis designed in this invention summarize, filter, and convert basic source data into a dataset of several easily identifiable and labeled parameter values ​​corresponding to those in the model across multiple nodes of each workpiece; simultaneously, a model is built for each parameter in multiple nodes of each workpiece through deep learning; then, the value of each parameter in multiple nodes of each workpiece after processing is compared with the parameter range in the model to determine whether the parameter value in the node where the workpiece is located is reasonable.

[0034] The technical solution of this invention can effectively perform intelligent crystal pulling based on big data and deep learning. It utilizes big data analysis and executes optimization schemes, and then organically combines big data and deep learning to improve crystal pulling quality and efficiency while reducing crystal pulling costs. Attached Figure Description

[0035] Figure 1 This is a flowchart of a crystal pulling method based on big data analysis according to an embodiment of the present invention;

[0036] Figure 2 This is a flowchart of the crystal pulling process according to an embodiment of the present invention;

[0037] Figure 3 This is a schematic diagram of the structure of a crystal pulling system according to an embodiment of the present invention.

[0038] In the picture: Detailed Implementation

[0039] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0040] This embodiment proposes a crystal pulling method based on big data analysis, such as... Figure 1 As shown, the steps include:

[0041] S1: Obtain the basic source data of multiple nodes for each workpiece in each single crystal furnace during the crystal pulling process.

[0042] Specifically, in the process of pulling each single crystal workpiece in each single crystal furnace, several nodes are involved. These nodes include at least a temperature stabilization node, a crystal pulling node, a shoulder forming node, a constant diameter node, and a termination node, which are set sequentially. The basic source data for each node includes production process data and / or raw material data and / or quality data.

[0043] The production process data includes equipment name, start and end time of each node, batch number, process mode, formula name, diameter measurement value, hot zone temperature value, main heater power measurement, bottom heater power measurement, and actual crystal pulling speed.

[0044] The raw material data includes the preparation date, batching sequence number, personnel shifts at each stage, furnace number, workpiece specifications, crucible type, crucible origin, weight of primary polycrystalline material, percentage of recycled material, and overall weight.

[0045] Quality data includes the single crystal number, length, weight, diameter, resistivity, lifespan, oxygen content, carbon content, and defects of each workpiece.

[0046] S2: Process the source data obtained in S1, filter and convert it into several parameters that are easy to identify and label in each node of each workpiece, and obtain a dataset of all parameter values ​​for each node of the workpiece.

[0047] Specifically, the basic data of each node is extracted, filtered, and converted into several parameters that are easy to identify and label in that node, so as to obtain a dataset of parameter values ​​that are easy to compare with the parameters in the standard module. In other words, the scattered, messy, and inconsistent source data in the input basic source data are integrated and then converted into a dataset of commonly used parameters in the workpiece process nodes, so as to provide a basis for subsequent parameter determination and analysis.

[0048] Furthermore, all parameters are established according to the production area, the start and end times of each node, the correlation data between upstream and downstream nodes, and the different functions in each node. All parameters for each node in each workpiece are configured and displayed on the terminal display of the single crystal furnace where the workpiece is being drawn, allowing personnel to monitor the changes in each single crystal furnace in real time.

[0049] S3: Build a model for each parameter in each node of each workpiece using deep learning.

[0050] Specifically, a deep learning approach is used to build a model for all parameters of each node of each workpiece, in order to monitor and analyze the nodes of all workpieces in the process of all furnaces, in order to obtain single crystal workpieces that meet the quality standards.

[0051] Furthermore, the parameter type of each node of each workpiece in the model corresponds to the parameter type of each node of each workpiece in step S2.

[0052] Furthermore, the model includes at least the model of the steady-temperature node, the model of the crystal-leading node, the model of the shoulder node, the model of the constant-diameter node, and the model of the terminal node.

[0053] S4: Compare the value of each parameter in each node of each workpiece in step S2 with the model in step S3 to determine whether the value of each parameter in the node of the workpiece is reasonable.

[0054] Specifically, if all parameter values ​​in the node where the workpiece is located are within the model range, then the operation of the next node continues.

[0055] If some parameter values ​​in the node where the workpiece is located are not within the model range, then the process will pause or adjust the relevant parameters in the corresponding node.

[0056] like Figure 2 The diagram shown is a flowchart of the crystal pulling process:

[0057] During the drawing process, the workpiece temperature stabilization data obtained from the dataset in the processing source data unit is matched with the data in the temperature stabilization node model. Temperature stabilization begins when the real-time data falls within the range specified in the temperature stabilization node model. Throughout the temperature stabilization process, the temperature stabilization node model automatically adjusts relevant parameter values ​​to maintain them within acceptable ranges until temperature stabilization ends. The temperature stabilization node model can also automatically determine whether the data at the end of the temperature stabilization node is suitable for crystal pulling; if so, it continues to the crystal pulling node process; if not, it adjusts to the step before temperature stabilization and starts from the beginning of the temperature stabilization node.

[0058] Upon entering the lead-in node process, the data from the workpiece lead-in obtained from the dataset is matched with the data in the lead-in node model. Lead-in begins when the real-time data falls within the range specified in the lead-in node model. Throughout the lead-in process, the lead-in node model automatically adjusts relevant parameter values ​​to maintain them within acceptable ranges until lead-in is complete. The lead-in node model can also automatically determine whether the data at the end of the lead-in node is suitable for shoulder formation; if so, it continues to the shoulder formation node process; if not, it backtracks to the step before temperature stabilization, proceeding step-by-step from the temperature stabilization node until the shoulder formation node process can begin.

[0059] In the shoulder-forming node process, the data from the workpiece shoulder-forming process obtained from the dataset is matched with the data in the shoulder-forming node model. Shoulder forming begins when the real-time data falls within the range specified in the shoulder-forming node model. Throughout the shoulder-forming process, the shoulder-forming node model automatically adjusts relevant parameter values ​​to maintain them within acceptable ranges until shoulder forming is complete. The shoulder-forming node model can also automatically determine whether the data at the end of the shoulder-forming node qualifies for shoulder rotation. If yes, it continues to the shoulder rotation node process; if not, it backtracks to the step before temperature stabilization and proceeds step-by-step from the start of the temperature stabilization node until it is ready to enter the shoulder rotation node process.

[0060] In the shoulder turning node process, the data from the workpiece shoulder turning obtained from the dataset is matched with the data in the shoulder turning node model. Shoulder turning begins when the real-time data falls within the range specified in the shoulder turning node model. Throughout the shoulder turning process, the shoulder turning node model automatically adjusts relevant parameter values ​​to ensure they are within acceptable ranges until the shoulder turning ends. The shoulder turning node model can also automatically determine whether the data at the end of the shoulder turning node is of equal diameter. If it is, it continues into the equal diameter node process; if not, it backtracks to the step before temperature stabilization and proceeds step-by-step from the start of the temperature stabilization node until it can enter the equal diameter node process.

[0061] Upon entering the equal diameter node process, the workpiece equal diameter data obtained from the dataset is matched with the data in the equal diameter node model. If the real-time data falls within the range specified in the equal diameter node model, the equal diameter process begins. Throughout the equal diameter process, the equal diameter node model automatically adjusts relevant parameter values ​​to ensure they remain within acceptable ranges until the equal diameter process ends. The equal diameter node model can also automatically determine whether the data at the end of the equal diameter node is acceptable for completion. If acceptable, it continues to the completion node process; otherwise, it backtracks to the steps before temperature stabilization and proceeds step-by-step from the start of the temperature stabilization node until the completion node process can begin.

[0062] In the final stage of the process, the data from the workpiece finishing stage obtained from the dataset is matched with the data in the final stage model. Finishing begins when the real-time data falls within the range specified in the final stage model. Throughout the finishing process, the final stage model automatically adjusts relevant parameters to ensure they remain within acceptable ranges until finishing is complete. The final stage model can also automatically determine whether the data at the end of the final stage allows for reprocessing or furnace shutdown. If it allows, the process continues to the reprocessing or shutdown stage; otherwise, it backtracks to the steps before temperature stabilization, proceeding step-by-step from the temperature stabilization stage until it can proceed to the reprocessing or shutdown stage.

[0063] A crystal pulling system, such as Figure 3 As shown, the system includes:

[0064] Source data acquisition unit: Used to acquire the basic source data of multiple nodes of each workpiece in each single crystal furnace during the crystal pulling process.

[0065] Source data processing unit: Used to process the acquired source data, filter and convert it into several parameters that are easy to identify and label in each node of each workpiece, and obtain a dataset of all parameter values ​​in each node of the workpiece.

[0066] Model building unit: used to build a model for each parameter in each node of each workpiece through deep learning.

[0067] Parameter Judgment Unit: Used to compare the value of each parameter in each node of each workpiece in the source data processing unit with the model in the model building unit to determine whether the value of each parameter in the node where the workpiece is located is reasonable.

[0068] The parameter determination unit includes the following steps: if all parameter values ​​in the node where the workpiece is located are within the model range, then the execution of the next node continues. If some parameter values ​​in the node where the workpiece is located are not within the model range, then the execution is paused or the relevant parameters in the corresponding node are adjusted.

[0069] Furthermore, the parameter types of each workpiece and each node in the source data unit are processed to correspond to the parameter types of each workpiece and each node in the model unit.

[0070] Furthermore, the parameters are established according to the production area, the start and end times of each node, the associated data between upstream and downstream nodes, and the different functions in each node.

[0071] Furthermore, all parameters of each node in each workpiece are configured and displayed on the terminal display of the single crystal furnace in which the workpiece is drawn.

[0072] Furthermore, the source data for each node in each workpiece includes production process data and / or raw material data and / or quality data.

[0073] Furthermore, each workpiece includes at least a temperature-stabilizing node, a crystal-leading node, a shoulder-forming node, a diameter-equalizing node, and a tail-ending node, and these nodes are set sequentially.

[0074] Furthermore, the model includes at least a model of a stable temperature node, a model of a crystal-leading node, a model of a shoulder-forming node, a model of a constant diameter node, and a model of a terminal node.

[0075] A computer device includes a memory and a processor; wherein the memory stores a computer program; and the processor is configured to execute the computer program and, when executing the computer program, cause the processor to perform the steps of the crystal pulling method as described in any of the preceding claims.

[0076] A computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to perform the steps of the crystal pulling method as described in any of the preceding claims.

[0077] 1. A crystal pulling method, system, computer equipment, and storage medium based on big data analysis, which summarizes, filters, and converts basic source data into a dataset of several easily identifiable and labeled parameter values ​​corresponding to those in the model across multiple nodes of each workpiece; simultaneously, a model is built for each parameter in multiple nodes of each workpiece through deep learning; then, the value of each parameter in multiple nodes of each workpiece after processing is compared with the parameter range in the model to determine whether the parameter value in the node of the workpiece is reasonable.

[0078] 2. The technical solution of this invention can effectively perform intelligent crystal pulling based on big data and deep learning. It utilizes big data analysis and executes optimization schemes, and then organically combines big data and deep learning to improve crystal pulling quality and efficiency, and reduce crystal pulling costs.

[0079] The embodiments of the present invention have been described in detail above. These descriptions are merely preferred embodiments and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A crystal pulling method based on big data analysis, characterized by the following steps: include: S1: Obtain the basic source data of multiple nodes for each workpiece in each single crystal furnace during the crystal pulling process; The source data for each node in each workpiece includes production process data and / or raw material data and / or quality data; S2: Process the acquired source data, filter and convert it into several parameters that are easy to identify and mark in each node of each workpiece, and obtain a dataset of the values ​​of all the parameters in each node of the workpiece; the parameters are established according to the production area, the start and end time of each node, the correlation data between the upper and lower nodes, and the different functions in each node, etc. S3: Establish a dynamic optimization model independently for each parameter in each node of each workpiece through deep learning; S4: Compare the value of each parameter in each node of each workpiece in S2 with the model in S3 to determine whether the value of each parameter in the node where the workpiece is located is reasonable; If any parameter value in the current node exceeds the model range, the process will be re-executed at the stable temperature node. If all parameter values ​​are within the model range, proceed to the next node.

2. The crystal pulling method based on big data analysis according to claim 1, characterized in that, All parameter types of each workpiece and each node in S2 correspond to all parameter types of each workpiece and each node in S3.

3. The crystal pulling method based on big data analysis according to claim 2, characterized in that... All parameters of each node in each workpiece are configured and displayed on the terminal display of the single crystal furnace in which the workpiece is drawn.

4. The crystal pulling method based on big data analysis according to any one of claims 1-3, characterized in that... Each workpiece includes at least a temperature-stabilizing node, a crystal-leading node, a shoulder-forming node, a constant-diameter node, and a tail-ending node, and the temperature-stabilizing node, the crystal-leading node, the shoulder-forming node, the constant-diameter node, and the tail-ending node are arranged sequentially.

5. The crystal pulling method based on big data analysis according to claim 4, characterized in that, The model includes at least a steady-temperature node model, a crystal-leading node model, a shoulder-forming node model, a constant-diameter node model, and a tail-end node model.

6. The crystal pulling method based on big data analysis according to claim 1, characterized in that, The process of retracing back to the stable temperature node and re-executing the entire process includes: starting from the stable temperature node and gradually running until entering the target node process; the dynamic optimization model adjusts parameters in real time based on historical time series data.

7. A crystal pulling system, characterized in that, The system includes: Source data acquisition unit: used to acquire basic source data of multiple nodes of each workpiece in each single crystal furnace in the crystal pulling process; the source data of each node in each workpiece includes production process data and / or raw material data and / or quality data; Source data processing unit: used to process the acquired source data, filter and convert it into several parameters that are easy to identify and mark in each node of each workpiece, and obtain a dataset of the values ​​of all the parameters in each node of the workpiece; wherein, the parameters are established according to the production area, the start and end time of each node, the correlation data between the upper and lower nodes, and the different functions in each node, etc. Model building unit: used to independently build a dynamic optimization model for each parameter in each node of each workpiece through deep learning; Parameter determination unit: used to compare the value of each parameter in each node of each workpiece in the processing source data unit with the model in the model building unit to determine whether the value of each parameter in the node where the workpiece is located is reasonable; The determination parameter unit is configured as follows: if any parameter value in the current node exceeds the model range, the control unit is triggered to backtrack to the stable temperature node and restart the entire process; if all parameter values ​​are within the model range, the next node continues to run.

8. The crystal pulling system according to claim 7, characterized in that, The parameter types of each workpiece and each node in the processing source data unit correspond to the parameter types of each workpiece and each node in the model building unit.

9. The crystal pulling system according to claim 7 or 8, characterized in that, All parameters of each node in each workpiece are configured and displayed on the terminal display of the single crystal furnace in which the workpiece is drawn.

10. The crystal pulling system according to claim 9, characterized in that, Each workpiece includes at least a temperature-stabilizing node, a crystal-leading node, a shoulder-forming node, a constant-diameter node, and a tail-ending node, and the temperature-stabilizing node, the crystal-leading node, the shoulder-forming node, the constant-diameter node, and the tail-ending node are arranged sequentially.

11. The crystal pulling system according to claim 10, characterized in that, The model includes at least a steady-temperature node model, a crystal-leading node model, a shoulder-forming node model, a constant-diameter node model, and a tail-end node model.

12. A computer device, characterized in that, It includes a memory and a processor; the memory stores a computer program; the processor is configured to execute the computer program, and when executing the computer program, cause the processor to perform the steps of the crystal pulling method as described in any one of claims 1-6.

13. A computer-readable storage medium, characterized in that, The device contains a computer program that, when executed by a processor, causes the processor to perform the steps of the crystal pulling method as described in any one of claims 1-6.

Images