Process parameter optimization method of solar cell diffusion process and related device

Abstract

The invention discloses a process parameter optimization method of a solar cell diffusion process and a related device. The method comprises the following steps: acquiring a historical process parameter data set and a historical sheet resistance quality detection data set corresponding to target diffusion equipment; preprocessing the historical process parameter data set and the historical sheet resistance quality detection data set to obtain a first process parameter data set and a first sheet resistance quality detection data set; determining a training data set and a test data set according to the first process parameter data set and the first sheet resistance quality detection data set; determining a target sheet resistance prediction model according to the training data set, the test data set and a preset sheet resistance prediction model; acquiring current process parameter data of the current batch of target diffusion equipment; and determining target process parameter data according to the current process parameter data and the target sheet resistance prediction model. By adopting the embodiment of the invention, the process parameters of the diffusion process can be accurately adjusted.

Description

Technical Field

[0001] This application relates to the field of solar cell technology, and in particular to a method and apparatus for optimizing process parameters in the diffusion process of solar cells. Background Technology

[0002] In the production of solar cells, the diffusion process is the core step in creating the PN junction and directly determines the photoelectric conversion efficiency of the solar cell. Sheet resistance is a key indicator for evaluating the quality of the diffusion process; its consistency and adherence to the target value are the core standards for measuring whether the process meets the standards.

[0003] In actual production, the sheet resistance value of the diffusion process needs to be kept highly stable to ensure consistent performance of solar cells. However, traditional process parameter adjustments rely entirely on engineers' experience and judgment, which cannot precisely regulate the diffusion process. This often leads to large fluctuations in sheet resistance yield and unstable production results, directly affecting the quality of solar cells.

[0004] Therefore, how to accurately adjust the process parameters of the diffusion process to improve the quality of solar cells has become an urgent problem to be solved. Summary of the Invention

[0005] This application provides a method and related apparatus for optimizing process parameters in the diffusion process of solar cells, which can achieve precise adjustment of process parameters in the diffusion process, thereby improving the quality of solar cells.

[0006] In a first aspect, embodiments of this application provide a method for optimizing process parameters in the diffusion process of a solar cell, including: Acquire historical process parameter datasets and historical sheet resistance quality test datasets corresponding to the target diffusion equipment; the target diffusion equipment is used to perform the solar cell diffusion process. Preprocessing operations are performed on the historical process parameter dataset and the historical sheet resistance quality detection dataset to obtain the first process parameter dataset and the first sheet resistance quality detection dataset. Based on the first process parameter dataset and the first sheet resistance quality detection dataset, determine the training dataset and the test dataset; Based on the training dataset, the test dataset, and the preset square resistance prediction model, determine the target square resistance prediction model; Obtain the current process parameter data of the target diffusion equipment for the current batch; Based on the current process parameter data and the target sheet resistance prediction model, the target process parameter data is determined.

[0007] Secondly, embodiments of this application provide a process parameter optimization device for the diffusion process of solar cells, comprising: an acquisition module and a determination module, wherein: The acquisition module is used to acquire historical process parameter datasets and historical sheet resistance quality test datasets corresponding to the target diffusion equipment; the target diffusion equipment is used to perform solar cell diffusion process; the historical process parameter datasets and the historical sheet resistance quality test datasets are preprocessed to obtain a first process parameter dataset and a first sheet resistance quality test dataset. The determining module is used to determine a training dataset and a test dataset based on the first process parameter dataset and the first sheet resistance quality detection dataset; and to determine a target sheet resistance prediction model based on the training dataset, the test dataset, and a preset sheet resistance prediction model. The acquisition module is also used to acquire the current process parameter data of the target diffusion equipment in the current batch; The determining module is further configured to determine the target process parameter data based on the current process parameter data and the target sheet resistance prediction model.

[0008] Thirdly, embodiments of this application provide an electronic device, including: a processor, a memory, a communication interface, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the processor, and the programs include instructions for performing the steps in the first aspect of embodiments of this application.

[0009] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program for electronic data interchange, wherein the computer program causes a computer to perform some or all of the steps described in the first aspect of embodiments of this application.

[0010] Fifthly, embodiments of this application provide a computer program product, wherein the computer program product includes a non-transitory computer-readable storage medium storing a computer program operable to cause a computer to perform some or all of the steps described in the first aspect of embodiments of this application. The computer program product may be a software installation package.

[0011] Implementing this application will have the following beneficial effects: As can be seen, the process parameter optimization method for the solar cell diffusion process described in this application, compared with the existing technology that relies on engineers' experience to adjust diffusion process parameters, adopts a technical route of using historical data of diffusion equipment to drive modeling and construct a sheet resistance prediction model to optimize parameters. This solves the problems of strong subjectivity in traditional process parameter adjustment, large fluctuations in sheet resistance pass rate, and inability to accurately match quality requirements. It has the advantages of strong equipment adaptability, precise quantitative control of parameters, and high quality stability, thereby achieving precise adjustment of diffusion process parameters and improving the quality of solar cells. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of this application or the background art, the accompanying drawings used in the embodiments of this application or the background art will be described below.

[0013] Figure 1 This is a schematic diagram of the structure of a target diffusion device provided in an embodiment of this application; Figure 2 This is a schematic diagram of the structure of a first furnace platform provided in an embodiment of this application; Figure 3 This is a flowchart of a method for optimizing process parameters in the diffusion process of a solar cell, provided in an embodiment of this application. Figure 4 This is a flowchart of a method for obtaining a fifth feature dataset provided in an embodiment of this application; Figure 5 This is a flowchart of a method for determining a target sheet resistance prediction model provided in an embodiment of this application; Figure 6 This is a flowchart of another method for determining a target sheet resistance prediction model provided in an embodiment of this application; Figure 7 This is a functional module block diagram of a process parameter optimization device for the diffusion process of solar cells provided in an embodiment of this application; Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0014] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0015] The terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or apparatuses.

[0016] It should be understood that the term "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this document indicates that the preceding and following related objects are in an "or" relationship. In the embodiments of this application, "multiple" refers to two or more.

[0017] In the embodiments of this application, "at least one item" or its similar expression refers to any combination of these items, including any combination of a single item or a plurality of items. "One or more" means one or more, while "multiple" means two or more. For example, "at least one item" of a, b, or c can represent the following seven cases: a, b, c; a and b; a and c; b and c; a, b, and c. Each of a, b, and c can be an element or a set containing one or more elements.

[0018] In this application, the term "connection" refers to various connection methods, such as direct connection or indirect connection, to achieve communication between devices. This application does not impose any limitations on this.

[0019] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0020] The electronic devices described in this application embodiment may include smartphones (such as Android phones, iOS phones, Windows Phones, etc.), tablet computers, PDAs, laptops, video matrices, monitoring platforms, mobile internet devices (MIDs), or wearable devices, etc. The above are merely examples and not exhaustive, and include but are not limited to the above devices.

[0021] Of course, the aforementioned electronic devices can also be servers, such as cloud servers.

[0022] The following describes the relevant content, concepts, meanings, technical issues, technical solutions, and beneficial effects involved in the embodiments of this application.

[0023] First, let me explain some of the technical terms or phrases used in this application: The diffusion process in solar cell manufacturing is a core step in forming the PN junction. Specifically, it involves incorporating dopants (such as phosphorus or boron) into the surface of a semiconductor silicon wafer through high-temperature diffusion. This alters the originally uniform doping concentration distribution on the silicon wafer, creating a semiconductor region on the surface with a conductivity type opposite to that of the substrate. This process constructs the PN junction, the core structure for photoelectric conversion in solar cells. This step is typically performed in a high-temperature tube diffusion furnace and mainly includes stages such as heating, isothermal diffusion, and cooling. It directly determines the upper limit of the photoelectric conversion efficiency of the solar cell.

[0024] Sheet resistance quality of solar cells: This is a core quantitative indicator for evaluating the processing effectiveness of the diffusion process, referring to the sheet resistance value and consistency of the doped region on the surface of the silicon wafer after diffusion. Its core evaluation dimensions include two points: first, the absolute sheet resistance value, i.e., the degree to which the measured sheet resistance value matches the preset target value; excessive deviation from the target value directly affects the built-in electric field strength of the PN junction; second, sheet resistance consistency, including the range of sheet resistance fluctuations between cells in the same batch and between different temperature zones of the same furnace tube. Poor consistency leads to inconsistent cell performance and reduces mass production yield.

[0025] Sheet resistance hit rate: This is a key statistical indicator for evaluating the quality control effectiveness of the solar cell diffusion process. It refers to the percentage of solar cells (or temperature zones) whose sheet resistance test values ​​fall within a preset acceptable range in a specific batch or time period, out of the total number of cells tested. The calculation logic is: Sheet resistance hit rate = (Number of acceptable sheet resistance values ​​ / Total number of cells tested) × 100%. This indicator directly reflects the accuracy of process parameter control. A higher hit rate indicates a more stable diffusion process and better performance consistency of the solar cells. It is a core basis for judging whether the process meets standards in the mass production stage.

[0026] PERC stands for Passivated Emitter and Rear Cell. It is a conventional high-efficiency battery technology developed from traditional aluminum back-surface solar cells. The core improvement is the addition of a passivation film (usually aluminum oxide or silicon nitride) and a local metal contact structure on the back of the cell.

[0027] TOPCon: Short for Tunnel Oxide Passivated Contact, it is a next-generation high-efficiency solar cell technology belonging to the passivated contact technology route and is currently one of the core directions for the upgrading of the photovoltaic industry.

[0028] Modbus protocol: an industrial-grade communication protocol specifically designed for data transmission between industrial devices (e.g., sensors, diffusion devices).

[0029] Kafka message queue: a high-throughput, highly reliable data transmission middleware.

[0030] PLC control system: Programmable Logic Controller system, is a digital computing and operating electronic system designed specifically for industrial environments.

[0031] MySQL is an open-source relational database.

[0032] As an important green energy source, solar energy has seen explosive growth in photovoltaic power generation in recent years. To continuously improve cell conversion efficiency, the photovoltaic industry's technology roadmap has been upgraded from PERC to TOPCon. This technology achieves efficient passivation on the back of the cell through a stacked structure composed of an ultra-thin SiO2 tunneling layer and a doped polycrystalline silicon layer, significantly improving the theoretical efficiency limit of the cell.

[0033] The TOPCon process mainly covers core steps such as high-temperature tubular diffusion, front and back coating, passivation treatment, and electrode fabrication. Among them, the core objective of the high-temperature tubular diffusion process is to form a built-in electric field in the PN junction between the N-type and P-type semiconductors, and the diffusion sheet resistance is a key indicator for measuring the quality of this process.

[0034] In actual production, the sheet resistance value of the diffusion process needs to be kept highly stable to ensure consistent performance of solar cells. However, traditional process parameter adjustments rely entirely on engineers' experience and judgment, which cannot precisely regulate the diffusion process. This often leads to large fluctuations in sheet resistance yield and unstable production results, directly affecting the quality of solar cells.

[0035] To address the aforementioned issues, this application provides a method and related apparatus for optimizing process parameters in the diffusion process of solar cells, which enables precise adjustment of process parameters in the diffusion process, thereby improving the quality of solar cells.

[0036] Please see Figure 1 , Figure 1 This is a schematic diagram of the structure of a target diffusion device provided in an embodiment of this application; the target diffusion device may include: multiple furnaces, data acquisition units, etc., which are not limited here; wherein: The furnace platform is the basic operating unit of the target diffusion equipment and also the carrier for hierarchical management of the equipment; please refer to [link / reference needed]. Figure 2 , Figure 2 This is a schematic diagram of the structure of a first furnace platform provided in an embodiment of this application. It can be seen that the first furnace platform may include multiple furnace tubes. The first furnace platform is any one of the multiple furnace platforms; that is, each of the multiple furnace platforms integrates multiple independent diffusion furnace tubes, possessing independent gas source distribution, temperature control scheduling, and process control capabilities. The core functions of the furnace platform include: This enables parallel processing of multiple furnace tubes, significantly improving the mass production efficiency of solar cells; The process execution status of the subordinate furnace tubes is centrally monitored to ensure the stability of the process parameters of each furnace tube; It facilitates production scheduling and allows for flexible allocation of different furnaces to execute differentiated diffusion process formulations according to demand.

[0037] The data acquisition unit is the information sensing core of the target diffusion device, and it can interact with various execution components of the device and external devices. The core functions of the data acquisition unit can include: Equipment information collection: Real-time recording of identification information such as furnace platform number, furnace tube number, and process batch number to achieve unique traceability of "equipment, batch, and product"; Process parameter acquisition: Key process data in each furnace tube are collected synchronously, including temperature in each temperature zone, diffusion source flow rate, oxygen / nitrogen flow rate, furnace tube pressure, pump speed, duration of each process stage, etc., to provide data support for subsequent processing; Quality data integration: A data interface is reserved to receive the batch temperature range sheet resistance test values ​​uploaded by the sheet resistance test instrument about 20 minutes after the diffusion process is completed, so as to achieve accurate matching of "process parameters and quality data"; Data preprocessing: The collected raw data is cleaned, denoised, and completed, and outliers are removed to ensure the validity and reliability of the data.

[0038] It should be explained that, in some embodiments, in addition to the core components mentioned above, the target diffusion equipment may also include a temperature control component (independently regulating the temperature of each temperature zone of the furnace tube), a gas path component (precisely controlling the gas source flow and furnace tube pressure), a main control module (coordinating the overall operation of the equipment and issuing process parameters), etc., which are not limited here; through the collaborative work of each component, the stable execution of the diffusion process is ensured.

[0039] Please see Figure 3 , Figure 3 This is a flowchart of a method for optimizing process parameters in the diffusion process of a solar cell, provided in an embodiment of this application. The method includes, but is not limited to, the following steps: S301. Obtain the historical process parameter dataset and historical sheet resistance quality test dataset corresponding to the target diffusion equipment; the target diffusion equipment is used to perform the solar cell diffusion process.

[0040] In this embodiment of the application, the process parameters may include at least one of the following: process batch number, process formula, temperature, diffusion source flow rate, oxygen content, nitrogen content, pump speed, pressure data, time, etc., which are not limited here; the sheet resistance quality detection data may include at least one of the following: process batch number, temperature zone number, sheet resistance detection value, sheet resistance hit rate, sheet resistance consistency, etc., which are not limited here.

[0041] The structure of the target diffusion device can be as follows: Figure 1 As shown.

[0042] In a specific embodiment, the data collection scope and related dimensions can be determined first, with the target diffusion device as the core, and the boundaries of the data to be collected can be clearly defined: Equipment dimension: Identify the furnace platform number and / or furnace tube number that need to be collected to obtain the equipment number data, and limit the collection to all process batches under the equipment corresponding to the equipment number data to exclude interference from other equipment data; Time dimension: Determine the time range of historical data based on modeling requirements (e.g., the last 3 months / the last 1000 batches); Association Dimension: Using the "process batch number" as the unique association key, we ensure that the process parameter data of each batch corresponds one-to-one with the sheet resistance quality inspection data.

[0043] Next, based on the data collection scope and related dimensions, the corresponding historical process parameter data and historical sheet resistance quality test data can be extracted from the preset database of the target diffusion equipment, thereby obtaining the historical process parameter dataset and the historical sheet resistance quality test dataset.

[0044] In some embodiments, the PLC control system, temperature control module, gas path sensor, and other hardware of the target diffusion equipment can be connected via the Modbus protocol to collect the process timing data of each batch in real time, including: furnace platform number, furnace tube number, process batch number, process formula, temperature timing data of each temperature zone, diffusion source flow timing data, oxygen / nitrogen flow timing data, pump speed timing data, furnace tube pressure timing data, duration of each process stage (heating / constant temperature / cooling), etc., which are not limited here; then, the collected real-time process data can be asynchronously transmitted using a Kafka message queue and the data can be transmitted to a preset database for storage to avoid data loss.

[0045] In some embodiments, approximately 20 minutes after the target diffusion device completes a diffusion process, the sheet resistance testing instrument is connected via the Modbus protocol to collect the sheet resistance quality testing data for that batch, including: process batch number, temperature zone number, and sheet resistance test value of the corresponding test piece for each temperature zone; similarly, the sheet resistance quality testing data can also be transmitted to a preset database for storage via a Kafka message queue and associated with the process parameter data of the same batch.

[0046] In some embodiments, persistent data storage can be achieved using Python technology in conjunction with a database (e.g., MySQL). Write a data cleaning and format conversion script in Python to standardize the raw data transmitted through the Kafka message queue: Remove outliers (e.g., out-of-range data caused by sensor malfunction) and fill in missing values ​​(e.g., missing parameters caused by brief communication interruptions). Organize the time series data in a structured manner according to "process stage and timestamp" and unify the data units (e.g., temperature unit is degrees Celsius, flow rate unit is standard cubic centimeters per minute). Construct a structured database table, divided into "Process Parameter Table" and "Sheet Resistance Quality Inspection Table": Process parameter table: The primary key can be "process batch number, furnace tube number", and the fields include furnace platform number, furnace tube number, process batch number, process formula, various time sequence parameter fields, process time, etc., which are not limited here; Sheet resistance quality inspection form: The primary key can be "process batch number, temperature zone number", and the fields include process batch number, furnace tube number, temperature zone number, sheet resistance test value, test time, etc., which are not limited here; The two tables are linked by an index that links the process batch number with the furnace tube number (or furnace platform number) to ensure data traceability.

[0047] S302. Perform preprocessing operations on the historical process parameter dataset and the historical sheet resistance quality detection dataset to obtain the first process parameter dataset and the first sheet resistance quality detection dataset.

[0048] In this embodiment of the application, the preprocessing operation may include at least one of the following: data integrity processing operation, data structured splitting operation, data cleaning operation, data standardization (or normalization) operation, etc., which are not limited here.

[0049] In a specific embodiment, the preprocessing operation may include data integrity processing and data structuring splitting. First, data integrity processing can be performed on the historical process parameter dataset and the historical sheet resistance quality inspection dataset. The goal of this operation is to remove missing samples that cannot be used for model training, as detailed below: Using "process batch number" as the association key, a two-way matching verification was performed on the historical process parameter dataset and the historical sheet resistance quality inspection dataset to remove missing samples: The first step, matching, involves iterating through all process batch numbers in the historical process parameter dataset and checking each batch against the historical sheet resistance quality inspection dataset for corresponding sheet resistance data. If a batch only has process parameter data but no corresponding sheet resistance inspection data (e.g., the sheet resistance testing instrument malfunctioned and did not collect data), then all process parameter data for that batch are directly removed. The second step, matching, involves iterating backwards through all process batch numbers in the historical sheet resistance quality inspection dataset and verifying whether the corresponding batch has complete process parameters in the historical process parameter dataset. If a batch only has sheet resistance inspection data but lacks process parameter data (e.g., sensor malfunction causing parameter non-collection), then all sheet resistance inspection data for that batch is removed. It should be explained that if only a few non-core fields (e.g., remarks) are missing, they can be supplemented by "mean of batches with the same formula from the same furnace" or "linear interpolation"; if core fields (e.g., temperature data) are missing, the sample data should be directly removed. Output results: After bidirectional rejection, a complete dataset with no missing data is obtained (each batch has complete process parameter data and sheet resistance test data).

[0050] Next, the basic dataset can be structurally split. Specifically, for the time-series data of process parameters in the basic dataset, it can be split according to the process stages of the diffusion process to standardize the data format. Process stage division: First, clarify the standard process stages of solar cell diffusion (e.g., heating stage, isothermal diffusion stage, cooling stage), and determine the time nodes for each stage; Time-series data splitting: Extract all process parameter time-series data for each batch, and according to the preset process stage time nodes, split the continuous time-series data of a single batch into "heating stage - parameter time-series data", "constant temperature stage - parameter time-series data", and "cooling stage - parameter time-series data". Data structuring: Add "process stage identifiers" (e.g., heating / constant temperature / cooling) to the split time-series data of each stage to form a process parameter data structure classified by process stage.

[0051] Next, the sheet resistance quality inspection data in the basic dataset can be split dimensionally according to the furnace tube temperature zone to match the spatial dimension of the process parameters: Temperature zone information association: Extract the "temperature zone number" field from each batch of sheet resistance test data (e.g., zone 1 of furnace tube a, zone 2 of furnace tube a, zone 3 of furnace tube a, etc.). Quality data splitting: The sheet resistance test data of the same batch is split into "each temperature zone - sheet resistance test value" according to the temperature zone number, that is, each temperature zone corresponds to a set of independent sheet resistance quality data. Data structuring: Add "temperature zone number identifier" to the sheet resistance data of each temperature zone after splitting to form a sheet resistance quality test data structure classified by temperature zone.

[0052] Finally, the process parameter data and sheet resistance quality inspection data, which have undergone missing data removal and structured splitting, can be integrated to obtain the final standardized dataset: Integrate all batches of process parameter data after being broken down by process stage to form the first process parameter dataset (data dimensions: process batch number - process stage - parameter type - time series value). Integrate all batches of sheet resistance test data after being separated by temperature zone to form the first sheet resistance quality test dataset (data dimension: process batch number - temperature zone number - sheet resistance test value). The two datasets still use "process batch number" as the core association key to ensure a one-to-one correspondence between process parameters and sheet resistance quality data, which can be directly used for subsequent model training.

[0053] It should be noted that preprocessing operations may also include other operations besides data integrity processing operations, which will not be elaborated here.

[0054] S303. Based on the first process parameter dataset and the first sheet resistance quality detection dataset, determine the training dataset and the test dataset.

[0055] In this embodiment of the application, data can be extracted from the first process parameter dataset and the first sheet resistance quality detection dataset to form a training dataset and a test dataset, respectively.

[0056] Optionally, the solar cell diffusion process includes c process stages; the target diffusion equipment includes d temperature zones; c and d are both integers greater than 1; the first process parameter dataset includes c process parameter data; each process parameter data corresponds to one process stage; the first sheet resistance quality detection dataset includes d sheet resistance quality detection data; each sheet resistance quality detection data corresponds to one temperature zone; step S303, determining the training dataset and test dataset based on the first process parameter dataset and the first sheet resistance quality detection dataset includes: S11. Determine the equipment information corresponding to each of the c process parameter data to obtain c equipment information; S12. Determine the first feature dataset based on the c process parameter data and the c equipment information; S13. Perform feature filtering and dimensionality reduction on the first feature dataset to obtain the second feature dataset; S14. Determine the third feature dataset based on the d sheet resistance quality detection data; S15. Determine the fourth feature dataset based on the third feature dataset and the second feature dataset; S16. Divide the fourth feature dataset according to a preset ratio to obtain the training dataset and the test dataset.

[0057] In this embodiment of the application, the preset ratio can be preset in advance or defaulted; each device information may include at least one of the following: device identification information (e.g., furnace tube number), device model information, device specification information (e.g., number of furnace tube temperature zones), device status information (e.g., device running time), etc., which are not limited here.

[0058] In a specific embodiment, firstly, all equipment information of the target diffusion equipment can be obtained. Then, for each process parameter data in c process parameter data, the bound process batch number and furnace tube number are extracted. All equipment information corresponding to the bound furnace tube number is extracted from the all equipment information to obtain one piece of equipment information, such as equipment model xx, number of temperature zones 5, running time 4800h, etc. In this way, c pieces of equipment information can be obtained. Then, based on the c process parameter data and c pieces of equipment information, a first feature dataset can be determined. Specifically, for each process parameter data, the data type contained therein can be determined to obtain multiple data types. Then, the process parameter data is split according to the multiple data types to obtain multiple subsets, each subset corresponding to a data type. Next, the mean of each subset can be calculated to obtain multiple means. Then, the first equipment information corresponding to the process parameter data in the c pieces of equipment information can be determined. The first equipment information is bound to these multiple means to obtain a feature data. In this way, c feature data, that is, the first feature dataset, can be obtained.

[0059] It should be noted that data types can include one of the following: temperature data type, oxygen data type, pressure data type, sheet resistance data type, etc., without limitation.

[0060] Then, feature filtering and dimensionality reduction can be performed on the first feature dataset to obtain the second feature dataset. Next, the third feature dataset can be determined based on d sheet resistance quality test data. Specifically, the average sheet resistance of each sheet resistance quality test data can be calculated to obtain d average sheet resistance values, which are then used as the third feature dataset. Then, the fourth feature dataset can be determined based on the third and second feature datasets. Specifically, the "process batch number" and "temperature zone number" of the two feature datasets can be set to the same data format, for example, the process batch number can be uniformly set to PC-YYMMDD-XXX, and the temperature zone number can be uniformly set to zone X. Next, data belonging to the same process batch and the same temperature zone in the third and second feature datasets can be merged to obtain the fourth feature dataset. Specifically, the merging operation can be as follows: The third feature dataset is used as a benchmark (square resistance features are the prediction target and have higher priority). Iterate through each record in the third feature dataset and extract its process batch number and temperature zone number; Based on the process batch number and temperature zone number, search for a completely matching record in the second feature dataset; The matched second feature data (e.g., average process parameters, equipment model, runtime) is directly appended to the corresponding row of the third feature data.

[0061] The third feature data after the final traversal is completed is the fourth feature dataset. Finally, the fourth feature dataset can be divided according to a preset ratio to obtain the training dataset and the test dataset. For example, the preset ratio can be 7:3, randomly selecting 70% of the data from the fourth feature dataset as the training dataset and the remaining 30% as the test dataset.

[0062] Thus, by binding process parameters and equipment information, a first feature dataset is generated; the first feature dataset is then filtered and dimensionality reduced to remove redundant interference, resulting in a refined second feature dataset; based on the sheet resistance detection data, a third feature dataset reflecting quality characteristics is constructed; then, the second feature dataset (reflecting process and equipment characteristics) and the third feature dataset are merged to form a fourth feature dataset with a causal relationship between process and quality; finally, the fourth feature dataset is divided into a training set and a test set according to a preset ratio to ensure that the model training has sufficient samples and to provide a reliable data foundation for the sheet resistance prediction model.

[0063] Optionally, step S13, which involves performing feature filtering and dimensionality reduction on the first feature dataset to obtain a second feature dataset, includes: S21. Determine the e data types corresponding to the first feature dataset; e is an integer greater than 1; S22. Determine the e first importance values ​​corresponding to the e data types; each first importance value corresponds to one data type; S23. Based on the preset correlation coefficient algorithm, the first feature dataset, and the first sheet resistance quality detection dataset, determine the correlation coefficients between the e data types and the sheet resistance quality detection data to obtain e correlation coefficients; each correlation coefficient corresponds to a data type. S24. Determine e second importance values ​​based on the e correlation coefficients and the e first importance values; S25. Based on the e second importance values ​​and the e data types, perform feature filtering on the first feature dataset to obtain the fifth feature dataset; S26. Perform dimensionality reduction processing on the fifth feature dataset according to the preset dimensionality reduction algorithm to obtain the second feature dataset.

[0064] In this embodiment of the application, both the preset correlation coefficient algorithm and the preset dimensionality reduction algorithm can be preset in advance or defaulted. The preset correlation coefficient algorithm may include one of the following: Pearson correlation coefficient algorithm, Spearman rank correlation coefficient algorithm, Kendall rank correlation coefficient algorithm, etc., which are not limited here. The preset dimensionality reduction algorithm may include at least one of the following: linear discriminant analysis (LDA) dimensionality reduction algorithm, principal component analysis (PCA) dimensionality reduction algorithm, etc., which are not limited here.

[0065] In a specific embodiment, field attributes can be extracted from the first feature dataset, all fields in the dataset can be listed, and distinctions can be made between identification categories (e.g., process batch number, furnace tube number), process categories (e.g., temperature, pressure), equipment categories (e.g., equipment model, equipment running time), quality categories (e.g., sheet resistance detection value), etc. Fields with the same attribute are grouped into one category, and the final number of categories is counted, which is e data types. Then, e first importance levels corresponding to the e data types can be determined. Specifically, a preset mapping relationship between data types and first importance levels can be stored in advance, and the e first importance levels corresponding to the e data types can be determined based on this mapping relationship.

[0066] Then, based on the preset correlation coefficient algorithm, the first feature dataset, and the first sheet resistance quality detection dataset, the correlation coefficients between e data types and the sheet resistance quality detection data can be determined, resulting in e correlation coefficients. For example, assuming the preset correlation coefficient algorithm is the Pearson correlation coefficient algorithm, corresponding data can be extracted from the first feature dataset based on the e data types to obtain e subset datasets. Then, sheet resistance quality indicators (e.g., sheet resistance detection values, sheet resistance average values) are extracted from the first sheet resistance quality detection dataset, and it is ensured that the two datasets are matched one-to-one by process batch number and temperature zone number. Then, the Pearson correlation coefficient algorithm is used to calculate the correlation coefficients between the e subset datasets and the sheet resistance quality indicators, thereby obtaining e correlation coefficients.

[0067] It should be noted that the correlation coefficient can range from -1 to 1; the importance can range from 0 to 1.

[0068] Next, based on the e correlation coefficients and e first importance values, e second importance values ​​are determined. Specifically, since the correlation coefficients and importance values ​​have different ranges, it is necessary to first normalize the e correlation coefficients and e first importance values ​​to the interval [0, 1], as follows: For the e correlation coefficients, take the absolute value and then perform Min-Max normalization to obtain e normalized correlation coefficients; For the e first importance values, directly perform Min-Max normalization to obtain e normalized first importance values; Then, the first weight corresponding to the correlation coefficient and the first weight corresponding to the first importance can be determined; wherein, the sum of the first weight and the second weight is 1; based on the first weight and the second weight, a weighted operation is performed on the e normalized correlation coefficients and the e normalized first importance, as follows: Second importance of target = Normalized correlation coefficient of target × First weight + Normalized first importance of target × Second weight; Wherein, the target normalized correlation coefficient is any one of the e normalized correlation coefficients; the target normalized first importance is the first importance corresponding to the target normalized correlation coefficient among the e normalized first importances; the target second importance is the second importance corresponding to the target normalized first importance; thus, by calculating e times, e second importances can be obtained.

[0069] In some embodiments, if more emphasis is placed on the strength of data association: the first weight can be 0.6, and the second weight can be 0.4; In some embodiments, if more emphasis is placed on the basic importance of features: the first weight can be 0.4 and the second weight can be 0.6.

[0070] Then, based on e secondary importance values ​​and e data types, feature filtering can be performed on the first feature dataset to obtain the fifth feature dataset; finally, a preset dimensionality reduction algorithm can be used to reduce the dimensionality of the fifth feature dataset to obtain the second feature dataset.

[0071] In some embodiments, the preset dimensionality reduction algorithm can be the LDA dimensionality reduction algorithm.

[0072] In some embodiments, the feature data of the j-th process batch in the second feature dataset is denoted as... , It can be as follows: ; in, This represents the m-th dimension data after dimensionality reduction.

[0073] Thus, by first dividing the data types of the first feature dataset and assigning them a basic first importance, and then combining the correlation coefficient algorithm to quantify the correlation strength between each type and the sheet resistance quality, a second importance that better meets the prediction requirements is obtained. Based on the second importance, high-value features are selected to generate the fifth feature dataset, irrelevant and redundant features are removed, and then the second feature dataset is further simplified by the dimensionality reduction algorithm. This not only retains the key process and equipment features for quality prediction, but also reduces the computational complexity of the model and avoids overfitting.

[0074] Optional, please refer to Figure 4 , Figure 4 This is a flowchart of a method for obtaining a fifth feature dataset according to an embodiment of this application. Step S25 involves filtering the first feature dataset based on the e second importance values ​​and the e data types to obtain the fifth feature dataset, including... Figure 4 The steps shown are as follows: S31. Determine the second importance values ​​among the e second importance values ​​that are greater than or equal to a preset importance value, to obtain f second importance values; f is a positive integer less than or equal to e; S32. Determine the f data types corresponding to the f second importance values ​​among the e data types; S33. Select data corresponding to the f data types from the first feature dataset to obtain the fifth feature dataset.

[0075] In this embodiment of the application, the preset importance can be preset in advance or defaulted.

[0076] In a specific embodiment, for each of the e second importance degrees, it can be compared with a preset importance degree to obtain f second importance degrees that are greater than or equal to the preset importance degree; then, for each of the f second importance degrees, its corresponding data type can be found from the e data types to obtain f data types; finally, data corresponding to the f data types can be filtered from the first feature dataset to obtain the fifth feature dataset. Specifically, the first feature dataset can be traversed, each record in the dataset can be checked one by one, and fields matching the f data types can be extracted; fields unrelated to the f data types can be removed, and only matching data can be retained; the filtered data can be integrated to obtain the fifth feature dataset.

[0077] In some embodiments, the feature data corresponding to the j-th process batch in the fifth feature dataset is denoted as... The details are as follows: ; in, arrive This represents m1 temperature feature data; arrive This represents the characteristic data of m² diffusion sources; arrive This represents m3 oxygen characteristic data; arrive This represents m4 pressure characteristic data; arrive This represents m5 pump speed characteristic data; arrive This represents m6 time-related feature data; This represents the equipment characteristic data (e.g., equipment number) for the j-th process batch; m1, m2, m3, m4, m5, and m6 are all positive integers. The sheet resistance quality detection data corresponding to the j-th process batch in the first feature dataset is denoted as . The details are as follows: ; in, This represents the temperature zone data (m0) in the sheet resistance quality inspection data of the j-th process batch, where the subscript m0 is the temperature zone number.

[0078] In this way, by selecting high-value features based on a quantification threshold (i.e., a preset importance) and filtering out weakly correlated redundant data, a fifth feature dataset is generated, which reduces computational costs and avoids model overfitting, effectively improving the accuracy and generalization ability of the square resistance prediction model.

[0079] S304. Determine the target sheet resistance prediction model based on the training dataset, the test dataset, and the preset sheet resistance prediction model.

[0080] In this embodiment of the application, the preset sheet resistance prediction model can be preset in advance or defaulted.

[0081] In a specific embodiment, a preset square resistance prediction model can be trained using a training dataset and a test dataset to obtain a target square resistance prediction model.

[0082] Optional, please refer to Figure 5 , Figure 5 This is a flowchart of a method for determining a target square resistance prediction model according to an embodiment of this application. Step S304 involves determining the target square resistance prediction model based on the training dataset, the test dataset, and a preset square resistance prediction model, including... Figure 5 The steps shown are as follows: S41. Input the training dataset into the preset sheet resistance prediction model to obtain the first sheet resistance prediction result; S42. Solve the model parameters of the preset square resistance prediction model according to the preset solution algorithm, the training dataset and the first square resistance prediction result to obtain the target model parameters; S43. Determine a reference resistance prediction model based on the target model parameters and the preset resistance prediction model; S44. Determine the target sheet resistance prediction model based on the preset nonlinear mapping function set, the test dataset, and the reference sheet resistance prediction model.

[0083] In this embodiment of the application, the preset solution algorithm and the preset nonlinear mapping function set can be preset in advance or defaulted. The preset solution algorithm may include one of the following: sequence minimum optimization algorithm, gradient descent algorithm, least squares method, Bayesian optimization algorithm, etc., which are not limited here.

[0084] In a specific embodiment, the training dataset can be input into a preset sheet resistance prediction model for prediction to obtain a first sheet resistance prediction result. Then, the model parameters of the preset sheet resistance prediction model are solved according to the preset solution algorithm, the training dataset, and the first sheet resistance prediction result to obtain the target model parameters. Specifically, the actual sheet resistance detection result corresponding to the training dataset can be obtained, and the preset solution algorithm can be called to iteratively calculate and adjust the model parameters with the goal of minimizing the error between the predicted value (i.e., the first sheet resistance prediction result) and the actual value (i.e., the actual sheet resistance detection result). When the error converges to a preset threshold (e.g., 1%) or reaches the preset iteration limit (e.g., 10000), the calculation is stopped and the final target model parameters are output.

[0085] In some embodiments, the preset square resistance prediction model can be a Support Vector Machine (SVM) model, and the preset solution algorithm can be a sequence minimum optimization algorithm. Specifically, the SVM model is as follows: ; in, This represents the model output, i.e., the sample data. Corresponding predicted resistance; This represents the model input, which is the i-th sample data in the training dataset; This represents a nonlinear mapping function used to map input features to a high-dimensional feature space; , These are all model hyperparameters. Represents the model weight vector. Indicates model bias; The transpose of the model weight vector; The SVM model must meet the following constraints: ; ; ; in, Representing data The corresponding actual value; " represents the predicted value of the SVM model; Indicates the tolerance range for prediction error; , All data The corresponding slack variables, Corresponding to the upper boundary constraint, Corresponding lower boundary constraint; The loss function L of the SVM model is shown below: ; in, This represents the preset penalty coefficient. This represents the total number of samples in the training dataset; By combining the constraints and loss function L of the SVM model, Lagrange multipliers are introduced, and the optimal parameters of the model are obtained using the sequence minimum optimization algorithm. , (i.e., target model parameters).

[0086] Next, a reference sheet resistance prediction model can be determined based on the target model parameters and the preset sheet resistance prediction model. Specifically, the model parameters of the preset sheet resistance prediction model are adjusted to the target model parameters to obtain the reference sheet resistance prediction model. Finally, the target sheet resistance prediction model can be determined based on the preset nonlinear mapping function set, the test dataset, and the reference sheet resistance prediction model.

[0087] Thus, the model is accurately constructed through hierarchical iterative optimization. First, the optimal model parameters are solved using the training dataset and preliminary prediction results to obtain a reference model that fits the training data, thus solving the problem of insufficient adaptability of the preset model parameters. Then, the model is tuned by combining a set of nonlinear mapping functions and a test set to adapt to the distribution of complex process data. The final target sheet resistance prediction model balances fitting accuracy and generalization ability, while reducing the computational difficulty of directly constructing complex models and improving development efficiency.

[0088] Optionally, the preset nonlinear mapping function set includes *a* nonlinear mapping functions; *a* is an integer greater than 1; please refer to [link to relevant documentation]. Figure 6 , Figure 6 This is a flowchart of another method for determining a target sheet resistance prediction model provided in an embodiment of this application. Step S44 involves determining the target sheet resistance prediction model based on a preset nonlinear mapping function set, the test dataset, and the reference sheet resistance prediction model, including... Figure 6 The steps shown are as follows: S51. Based on the a nonlinear mapping functions and the reference sheet resistance prediction model, determine a sheet resistance prediction models; S52. Input the test dataset into each of the a-square resistance prediction models to obtain a-square resistance prediction results; S53. Determine the a error index values ​​corresponding to the a sheet resistance prediction results; each error index value corresponds to a sheet resistance prediction result. S54. Determine the minimum value among the a error index values, and determine the nonlinear mapping function corresponding to the minimum value among the a nonlinear mapping functions as the target nonlinear mapping function; S55. Determine the target sheet resistance prediction model based on the target nonlinear mapping function and the reference sheet resistance prediction model.

[0089] In the embodiments of this application, each nonlinear mapping function may include one of the following: radial basis kernel function, polynomial kernel function, sigmoid kernel function, etc., which are not limited here.

[0090] In a specific embodiment, *a* nonlinear mapping functions can be substituted into the reference sheet resistance prediction model to obtain *a* sheet resistance prediction models. Then, the test dataset can be input into each of the *a* sheet resistance prediction models to obtain *a* sheet resistance prediction results. Specifically, for each sheet resistance prediction model, by inputting the test dataset into the model, a sheet resistance prediction result can be obtained, thus obtaining *a* sheet resistance prediction results. Next, *a* error index values ​​corresponding to the *a* sheet resistance prediction results can be determined. Specifically, the actual sheet resistance detection results corresponding to the test dataset can be obtained, and the errors between the *a* sheet resistance prediction results and the actual sheet resistance detection results corresponding to the test dataset can be calculated to obtain *a* error index values. The error index can include one of the following: mean absolute error, mean square error, mean absolute percentage error, etc., which are not limited here.

[0091] To illustrate, suppose a=2, the number of samples in the test dataset is 3, and the error metric is mean absolute error, as follows: The predicted sheet resistance values ​​for the first sheet resistance prediction model are: 19.5, 23.2, and 20.3. The predicted sheet resistance values ​​for the second sheet resistance prediction model are: 20.1, 18.5, and 22.3. The actual sheet resistance detection results corresponding to the test dataset are: 20.0, 23.0, and 21.0. It can be seen that the first error index value (i.e., mean absolute error) of the first sheet resistance prediction result is as follows: (|19.5-20.0|+|23.2-23.0|+|20.3-21.0|) / 3=(0.5+0.2+0.7) / 3≈0.47; The second error index value (i.e., mean absolute error) of the second sheet resistance prediction result is as follows: (|20.1-20.0|+|18.5-23.0|+|22.3-21.0|) / 3=(0.1+4.5+1.3) / 3≈1.97; The first error index value of 0.47 is significantly smaller than the second error index value of 1.97. Therefore, it can be determined that the first square resistance prediction model has a smaller error and higher prediction accuracy.

[0092] Furthermore, the minimum value among the a error index values ​​can be found, and the nonlinear mapping function corresponding to the minimum value among the a nonlinear mapping functions can be determined as the target nonlinear mapping function. Finally, the target nonlinear mapping function can be substituted into the reference sheet resistance prediction model to obtain the target sheet resistance prediction model.

[0093] In this way, by constructing multiple candidate models and verifying the error index values ​​of each candidate model through a test set, the mapping function corresponding to the minimum error is selected to construct the target square resistance prediction model. This method quantifies the error and selects the best to avoid subjective bias, ensuring that the accuracy of the target square resistance prediction model is the optimal among the candidate models, which can significantly improve the accuracy and reliability of square resistance prediction.

[0094] S305. Obtain the current process parameter data of the target diffusion equipment in the current batch.

[0095] In this embodiment, the batch number corresponding to the current batch can be obtained first, and the process parameters can be collected from the preset database of the target diffusion device based on the batch number to obtain the current process parameter data.

[0096] S306. Determine the target process parameter data based on the current process parameter data and the target sheet resistance prediction model.

[0097] In this embodiment of the application, the target process parameter data can be obtained by analyzing the current process parameter data using a target sheet resistance prediction model.

[0098] Optionally, step S306, determining the target process parameter data based on the current process parameter data and the target sheet resistance prediction model, includes: S61. Obtain the temperature dataset of the target diffusion device in the previous batch of the current batch to obtain the first temperature dataset; S62. Adjust the first temperature dataset according to the preset temperature step size to obtain b temperature datasets; b is an integer greater than 1. S63. Combine each of the b temperature datasets with the current process parameter data to obtain b sets of process parameter datasets; S64. Input the b sets of process parameter datasets into the target sheet resistance prediction model to obtain b sheet resistance prediction results; S65. Determine the b square resistance hit rates corresponding to the b square resistance prediction results; each square resistance hit rate corresponds to one square resistance prediction result. S66. Determine the maximum value among the b sheet resistance hit rates, and determine the temperature dataset corresponding to the maximum value in the b temperature datasets as the target process parameter data.

[0099] In this embodiment of the application, the preset temperature step size can be preset in advance or defaulted. For example, the preset temperature step size can be 1 degree Celsius.

[0100] In a specific embodiment, the batch number of the current batch can be obtained first. Then, the batch number of the previous batch can be determined according to the batch number naming rule and the batch number of the current batch. For example, assuming the batch number naming rule is "batch numbers are arranged according to a continuous incrementing rule", the batch number of the previous batch can be obtained by subtracting one from the batch number. Next, the corresponding temperature dataset can be extracted from the preset database according to the batch number of the previous batch to obtain the first temperature dataset. Then, the first temperature dataset can be adjusted according to the preset temperature step size to obtain b temperature datasets. Specifically, all temperatures in the first temperature dataset can be increased (or decreased) by a preset temperature step size to obtain the second temperature dataset. Then, all temperatures in the second temperature dataset can be increased (or decreased) by a preset temperature step size to obtain the third temperature dataset. In this way, b temperature datasets can be obtained by adjusting b times.

[0101] Next, for each of the b temperature datasets, it can be combined with the current process parameter data to obtain a set of process parameter datasets, thus obtaining b sets of process parameter datasets. Then, each of the b sets of process parameter datasets can be input into the target sheet resistance prediction model for prediction, resulting in b sheet resistance prediction results. Furthermore, the b sheet resistance hit rates corresponding to the b sheet resistance prediction results can be determined. Specifically, for each sheet resistance prediction result, the predicted value is compared with the corresponding true value, and the relative error between the single set of predicted values ​​and the true values ​​is calculated. Relative error = |predicted value - actual value| / actual value × 100%; If the relative error is less than or equal to the preset threshold, it is considered a hit; otherwise, it is considered a miss. Calculate the number of hits and the total number of test samples (i.e., the number of predicted sheet resistances in the sheet resistance prediction results), and substitute them into the following formula: Square block hit rate = (number of hits / total number of test samples) × 100%; Based on the above formula, the hit rate of b square blocks can be obtained; Finally, we can find the maximum value among these b sheet resistance hit rates, and determine the temperature dataset corresponding to this maximum value in the b temperature datasets as the target process parameter data.

[0102] In some embodiments, the first temperature dataset can be denoted as: ;in, Indicates the temperature parameters of temperature range m1, superscript This indicates that the parameter is to be optimized, and the subscript m1 represents the temperature zone number; the parameter mesh is divided into separate grids for the first temperature dataset: For the temperature parameters of each temperature zone, the current temperature reference value is increased by q℃ and decreased by q℃ respectively with a preset temperature step size, ultimately forming m1. 2q / T A set of different temperature parameter combinations, where q is a natural number and T represents the preset temperature step size.

[0103] Taking m1=2 (2 temperature zones), q=2, and T=1 as an example, assume the current temperature reference values ​​for the 2 temperature zones are: Temperature Zone 1 baseline value: =800 (°C); Reference value for temperature zone 2: =850 (°C); Each temperature zone is adjusted independently, with q=2, meaning that the temperature is decreased by 2℃ and increased by 2℃ respectively with a preset temperature step size T, resulting in four adjusted temperatures for each temperature zone: Temperature adjustment result for Zone 1: 800 2 = 798 (℃), 800 - 1 = 799 (℃), 800 + 1 = 801 (℃), 800 + 2 = 802 (℃); Temperature adjustment result for Zone 2: 850 2 = 848 (℃), 850 1 = 849 (℃), 850 + 1 = 851 (℃), 850 + 2 = 852 (℃); By pairing the adjusted temperature parameters of the two temperature zones together, we obtain m1. 2q / T =2 4 =16 (groups) of different temperature parameter combinations.

[0104] m1 2q / T Each set of temperature parameter combinations is combined with the current process parameter data to obtain m1. 2q / T Set up a process parameter dataset; put m1 2q / T The datasets of process parameters are input into the target sheet resistance prediction model for prediction, and m1 is obtained. 2q / T The prediction results for each square resistance are obtained; based on the rules of square resistance hit rate, m1 is calculated. 2q / T m1 corresponding to the predicted resistance 2q / T The hit rate of each method; find this m1 2q / T The maximum value among the various blocking rates, and this maximum value in m1 2q / T The temperature parameter combination corresponding to the group of temperature parameter combinations is taken as the optimal temperature parameter combination (i.e., the target process parameter data).

[0105] As can be seen, the process parameter optimization method for the solar cell diffusion process described in this application, compared with the existing technology that relies on engineers' experience to adjust diffusion process parameters, adopts a technical route of using historical data of diffusion equipment to drive modeling and construct a sheet resistance prediction model to optimize parameters. This solves the problems of strong subjectivity in traditional process parameter adjustment, large fluctuations in sheet resistance pass rate, and inability to accurately match quality requirements. It has the advantages of strong equipment adaptability, precise quantitative control of parameters, and high quality stability, thereby achieving precise adjustment of diffusion process parameters and improving the quality of solar cells.

[0106] Please see Figure 7 , Figure 7 This is a functional block diagram of a process parameter optimization device 700 for the diffusion process of a solar cell, provided in an embodiment of this application. The process parameter optimization device 700 includes: an acquisition module 701 and a determination module 702, wherein: The acquisition module 701 is used to acquire the historical process parameter dataset and the historical sheet resistance quality test dataset corresponding to the target diffusion equipment; the target diffusion equipment is used to perform the solar cell diffusion process; the historical process parameter dataset and the historical sheet resistance quality test dataset are preprocessed to obtain the first process parameter dataset and the first sheet resistance quality test dataset. The determining module 702 is used to determine a training dataset and a test dataset based on the first process parameter dataset and the first sheet resistance quality detection dataset; and to determine a target sheet resistance prediction model based on the training dataset, the test dataset, and a preset sheet resistance prediction model. The acquisition module 701 is also used to acquire the current process parameter data of the target diffusion equipment in the current batch; The determining module 702 is further configured to determine the target process parameter data based on the current process parameter data and the target sheet resistance prediction model.

[0107] In specific implementations, the process parameter optimization device 700 for the solar cell diffusion process described in the embodiments of the present invention can also execute other implementations described in the process parameter optimization method for the solar cell diffusion process provided in the embodiments of the present invention, which will not be repeated here.

[0108] Please see Figure 8 , Figure 8This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device may include a processor, a memory, a communication interface, and one or more programs. The processor, memory, and communication interface can be interconnected via a bus. The one or more programs are stored in the memory and configured to be executed by the processor. In this embodiment, the programs include instructions for performing the following steps: Acquire historical process parameter datasets and historical sheet resistance quality test datasets corresponding to the target diffusion equipment; the target diffusion equipment is used to perform the solar cell diffusion process. Preprocessing operations are performed on the historical process parameter dataset and the historical sheet resistance quality detection dataset to obtain the first process parameter dataset and the first sheet resistance quality detection dataset. Based on the first process parameter dataset and the first sheet resistance quality detection dataset, determine the training dataset and the test dataset; Based on the training dataset, the test dataset, and the preset square resistance prediction model, determine the target square resistance prediction model; Obtain the current process parameter data of the target diffusion equipment for the current batch; Based on the current process parameter data and the target sheet resistance prediction model, the target process parameter data is determined.

[0109] In specific implementations, the electronic devices described in the embodiments of the present invention may also execute other implementation methods described in the above method embodiments, which will not be repeated here.

[0110] This application also provides a computer-readable storage medium storing a computer program for electronic data interchange, which causes a computer to perform some or all of the steps of any of the methods described in the above method embodiments, wherein the computer includes an electronic device.

[0111] This application also provides a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program operable to cause a computer to perform some or all of the steps of any of the methods described in the above method embodiments. The computer program product may be a software installation package, and the computer may include an electronic device.

[0112] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0113] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0114] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical or other forms.

[0115] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. This program can be stored in a computer-readable storage medium, and when executed, it can include the processes described in the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM or random access memory (RAM), magnetic disks, or optical disks.

[0116] The steps of the methods or algorithms described in the embodiments of this application can be implemented in hardware or by a processor executing software instructions. The software instructions can consist of corresponding software modules, which can be stored in RAM, flash memory, ROM, EPROM, electrically erasable programmable read-only memory (EEPROM), registers, hard disk, portable hard disk, read-only optical disk (CD-ROM), or any other form of storage medium well known in the art. An exemplary storage medium is coupled to a processor, enabling the processor to read information from and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and storage medium can reside in an ASIC. Furthermore, the ASIC can reside in a terminal device or management device. Alternatively, the processor and storage medium can exist as discrete components in the terminal device or management device.

[0117] Those skilled in the art will recognize that, in one or more of the examples above, the functions described in the embodiments of this application can be implemented, in whole or in part, by software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, in the form of a computer program product. This computer program product includes one or more computer instructions. When these computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated.

[0118] The aforementioned computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media.

[0119] The available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., digital video discs (DVDs)), or semiconductor media (e.g., solid-state disks (SSDs)).

[0120] The modules / units included in the various devices and products described in the above embodiments can be software modules / units, hardware modules / units, or a combination of both. For example, for devices and products applied to or integrated into a chip, all modules / units can be implemented using hardware methods such as circuits, or at least some modules / units can be implemented using software programs that run on a processor integrated within the chip, while the remaining (if any) modules / units can be implemented using hardware methods such as circuits. For devices and products applied to or integrated into a chip module, all modules / units can be implemented using hardware methods such as circuits. Different modules / units can be located in the same component (e.g., chip, circuit module, etc.) or different components of the chip module, or at least some modules / units can be implemented using hardware methods such as circuits. The implementation is achieved through a software program that runs on the processor integrated within the chip module. The remaining modules / units (if any) can be implemented using hardware methods such as circuits. For various devices and products applied to or integrated into terminal equipment, each of their modules / units can be implemented using hardware methods such as circuits. Different modules / units can be located in the same component (e.g., chip, circuit module, etc.) or different components within the terminal equipment. Alternatively, at least some modules / units can be implemented through a software program that runs on the processor integrated within the terminal equipment, while the remaining modules / units (if any) can be implemented using hardware methods such as circuits.

[0121] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the embodiments of this application. It should be understood that the above descriptions are merely specific embodiments of the embodiments of this application and are not intended to limit the protection scope of the embodiments of this application. Any modifications, equivalent substitutions, improvements, etc., made on the basis of the technical solutions of the embodiments of this application should be included within the protection scope of the embodiments of this application.

Claims

1. A method for optimizing process parameters in the diffusion process of solar cells, characterized in that, include: Obtain the historical process parameter dataset and historical sheet resistance quality test dataset corresponding to the target diffusion equipment; The target diffusion device is used to perform the solar cell diffusion process; Preprocessing operations are performed on the historical process parameter dataset and the historical sheet resistance quality detection dataset to obtain the first process parameter dataset and the first sheet resistance quality detection dataset. Based on the first process parameter dataset and the first sheet resistance quality detection dataset, determine the training dataset and the test dataset; Based on the training dataset, the test dataset, and the preset square resistance prediction model, determine the target square resistance prediction model; Obtain the current process parameter data of the target diffusion equipment for the current batch; Based on the current process parameter data and the target sheet resistance prediction model, the target process parameter data is determined; The step of determining the target process parameter data based on the current process parameter data and the target sheet resistance prediction model includes: Obtain the temperature dataset of the target diffusion device in the previous batch of the current batch to obtain the first temperature dataset; Based on a preset temperature step size, the first temperature dataset is adjusted to obtain b temperature datasets; b is an integer greater than 1. Each of the b temperature datasets is combined with the current process parameter data to obtain b sets of process parameter datasets; The target process parameter data is determined based on the b-group process parameter dataset and the target sheet resistance prediction model.

2. The method as described in claim 1, characterized in that, The step of determining the target resistance prediction model based on the training dataset, the test dataset, and the preset resistance prediction model includes: The training dataset is input into the preset sheet resistance prediction model to obtain the first sheet resistance prediction result; The model parameters of the preset square resistance prediction model are solved according to the preset solution algorithm, the training dataset, and the first square resistance prediction result to obtain the target model parameters. Based on the target model parameters and the preset sheet resistance prediction model, a reference sheet resistance prediction model is determined; The target sheet resistance prediction model is determined based on the preset nonlinear mapping function set, the test dataset, and the reference sheet resistance prediction model.

3. The method as described in claim 2, characterized in that, The preset nonlinear mapping function set includes a nonlinear mapping functions; a is an integer greater than 1; The step of determining the target sheet resistance prediction model based on the preset nonlinear mapping function set, the test dataset, and the reference sheet resistance prediction model includes: Based on the a nonlinear mapping functions and the reference sheet resistance prediction model, determine a sheet resistance prediction models; The test dataset is input into each of the a-sheet resistance prediction models to obtain a-sheet resistance prediction results; Determine a error index values ​​corresponding to the a sheet resistance prediction results; each error index value corresponds to one sheet resistance prediction result. Determine the minimum value among the a error index values, and determine the nonlinear mapping function corresponding to the minimum value among the a nonlinear mapping functions as the target nonlinear mapping function; The target sheet resistance prediction model is determined based on the target nonlinear mapping function and the reference sheet resistance prediction model.

4. The method according to any one of claims 1-3, characterized in that, The step of determining the target process parameter data based on the b-group process parameter dataset and the target sheet resistance prediction model includes: The b sets of process parameter datasets are input into the target sheet resistance prediction model to obtain b sheet resistance prediction results; Determine the b square resistance prediction results and their corresponding b square resistance hit rates; each square resistance hit rate corresponds to one square resistance prediction result. Determine the maximum value among the b sheet resistance hit rates, and determine the temperature dataset corresponding to the maximum value in the b temperature datasets as the target process parameter data.

5. The method according to any one of claims 1-3, characterized in that, The solar cell diffusion process includes c process stages; the target diffusion equipment includes d temperature zones; c and d are both integers greater than 1; the first process parameter dataset includes c process parameter data. Each process parameter data corresponds to a process stage; the first sheet resistance quality detection dataset includes d sheet resistance quality detection data. Each resistance quality test data corresponds to a temperature zone; The step of determining the training dataset and the test dataset based on the first process parameter dataset and the first sheet resistance quality detection dataset includes: Determine the equipment information corresponding to each of the c process parameter data to obtain c equipment information; Based on the c process parameter data and the c equipment information, a first feature dataset is determined; The first feature dataset is subjected to feature filtering and dimensionality reduction to obtain the second feature dataset. Based on the d sheet resistance quality detection data, the third feature dataset is determined; Based on the third feature dataset and the second feature dataset, a fourth feature dataset is determined; The fourth feature dataset is divided according to a preset ratio to obtain the training dataset and the test dataset.

6. The method as described in claim 5, characterized in that, The process of performing feature filtering and dimensionality reduction on the first feature dataset to obtain the second feature dataset includes: Determine the e data types corresponding to the first feature dataset; e is an integer greater than 1; Determine the e first importance values ​​corresponding to the e data types; each first importance value corresponds to one data type; Based on the preset correlation coefficient algorithm, the first feature dataset, and the first sheet resistance quality detection dataset, the correlation coefficients between the e data types and the sheet resistance quality detection data are determined, resulting in e correlation coefficients; each correlation coefficient corresponds to a data type. Based on the e correlation coefficients and the e first importance values, determine the e second importance values; Based on the e second importance values ​​and the e data types, feature filtering is performed on the first feature dataset to obtain the fifth feature dataset; The fifth feature dataset is subjected to dimensionality reduction processing according to a preset dimensionality reduction algorithm to obtain the second feature dataset.

7. The method as described in claim 6, characterized in that, The step of performing feature filtering on the first feature dataset based on the e second importance values ​​and the e data types to obtain the fifth feature dataset includes: Determine the second importance values ​​among the e second importance values ​​that are greater than or equal to a preset importance value to obtain f second importance values; f is a positive integer less than or equal to e. Determine the f data types corresponding to the f second importance values ​​among the e data types; The fifth feature dataset is obtained by filtering out the data corresponding to the f data types from the first feature dataset.

8. A device for optimizing process parameters in the diffusion process of solar cells, characterized in that, include: The module to obtain information and the module to determine information, wherein: The acquisition module is used to acquire historical process parameter datasets and historical sheet resistance quality test datasets corresponding to the target diffusion equipment; the target diffusion equipment is used to perform solar cell diffusion process; the historical process parameter datasets and the historical sheet resistance quality test datasets are preprocessed to obtain a first process parameter dataset and a first sheet resistance quality test dataset. The determining module is used to determine a training dataset and a test dataset based on the first process parameter dataset and the first sheet resistance quality detection dataset; and to determine a target sheet resistance prediction model based on the training dataset, the test dataset, and a preset sheet resistance prediction model. The acquisition module is also used to acquire the current process parameter data of the target diffusion equipment in the current batch; The determining module is further configured to determine the target process parameter data based on the current process parameter data and the target sheet resistance prediction model.

9. An electronic device, characterized in that, include: Processor, memory, communication interface, and one or more programs; The one or more programs are stored in the memory and configured to be executed by the processor, the programs including instructions for performing the steps of the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, A computer program for storing electronic data interchange, wherein the computer program causes a computer to perform the method as described in any one of claims 1-7.

Images