Cold hydrogenation process parameter prediction method, optimization method, device and equipment

Abstract

The invention discloses a cold hydrogenation process parameter prediction method, an optimization method, a cold hydrogenation process parameter prediction device and cold hydrogenation process parameter optimization equipment. The cold hydrogenation process parameter prediction method comprises the following steps: acquiring initial process parameters of a cold hydrogenation process flow; obtaining a pre-created cold hydrogenation process parameter composite model, wherein the cold hydrogenation process parameter composite model comprises a front-section mechanism model, a reactor big data model and a rear-section mechanism model; and based on the initial process parameters, simulating the whole cold hydrogenation process flow by utilizing a cold hydrogenation process parameter composite model to obtain prediction parameters of the cold hydrogenation process flow. According to the technical scheme, the whole technological process can be simulated and predicted by obtaining the initial technological parameters and utilizing the composite model composed of the front-section mechanism model, the reactor big data model and the rear-section mechanism model, and accurate prediction and analysis of the technological performance are achieved.

Description

Technical Field

[0001] This invention belongs to the field of cold hydrogenation technology, specifically relating to a method, optimization method, apparatus and equipment for predicting cold hydrogenation process parameters. Background Technology

[0002] Hydrogenation technology is an effective way to process polysilicon byproducts. The mainstream technology is cold hydrogenation. The cold hydrogenation process involves introducing silicon powder, catalyst, silicon tetrachloride, and hydrogen into a fluidized bed reactor. Under specific temperature and pressure conditions, the silicon is converted into trichlorosilane, which is then purified through multi-stage distillation. During this production process, continuous optimization of process parameters is necessary to achieve higher conversion rates with lower energy consumption.

[0003] Silicon chlorohydrin (SiHCl3) is an important intermediate in polysilicon production, and its conversion rate has a critical impact on production efficiency and product quality. However, due to limitations in current technology, real-time measurement of the SiHCl3 conversion rate is not possible, resulting in a lack of timeliness in adjusting process parameters and an inability to quickly respond to production changes. Furthermore, methods based on pure mechanistic modeling or pure big data modeling suffer from poor convergence and low prediction accuracy when predicting component yields, affecting the reliability of the prediction results and failing to effectively guide the optimization and adjustment of process parameters.

[0004] Therefore, existing technologies cannot accurately predict the process parameters of cold hydrogenation processes. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to address the above-mentioned shortcomings of the prior art by providing a method, optimization method, apparatus and equipment for predicting cold hydrogenation process parameters. Using this method can improve the accuracy of cold hydrogenation process parameter prediction.

[0006] In a first aspect, embodiments of the present invention provide a method for predicting cold hydrogenation process parameters, the method comprising:

[0007] Obtain the initial process parameters of the cold hydrogenation process, which are the process conditions and operating parameters that are pre-set at the start of the cold hydrogenation process and used to simulate and predict the performance of the process.

[0008] Obtain a pre-created composite model of cold hydrogenation process parameters, which includes a front-end mechanism model, a reactor big data model, and a back-end mechanism model;

[0009] Based on the initial process parameters, the entire cold hydrogenation process is simulated using a composite model of cold hydrogenation process parameters to obtain the predicted parameters of the cold hydrogenation process.

[0010] Preferably, the method further includes: collecting historical process parameters and experimental analysis data of the cold hydrogenation process;

[0011] Based on the historical process parameters and experimental analysis data, and based on chemical thermodynamics, reaction mechanism, distillation theory and fluidized bed reaction kinetics, an initial mechanism model for the entire cold hydrogenation process is established. The initial mechanism model includes at least a heat exchanger module, a fluidized bed reactor module, a quench tower module, a cooler module, a flash tank module, a distillation tower module, a hydrogen compressor module, and a pump module.

[0012] Correlation analysis was performed on the historical process parameters to identify key reactor parameters, which are those that affect the performance of the fluidized bed reactor.

[0013] Based on the key reactor parameters in the historical process parameters, a big data model of the reactor is obtained by training a neural network.

[0014] By replacing the fluidized bed reactor module in the initial mechanistic model with the reactor big data model, the composite model of the cold hydrogenation process parameters is obtained.

[0015] The front-end mechanism model includes the heat exchanger module, and the back-end mechanism model includes the quench tower module, cooler module, flash tank module, distillation tower module, hydrogen compressor module, and pump module.

[0016] Preferably, the step of performing correlation analysis on the historical process parameters to screen out key reactor parameters specifically includes:

[0017] The collected historical process parameters were analyzed using the Pearson correlation coefficient algorithm to determine the key parameters of the reactor.

[0018] The expression for the Pearson correlation coefficient analysis algorithm is as follows (1):

[0019]

[0020] Where r represents the correlation result, and n, xi, x, and S are the correlation coefficients. x y represents the number of sample operations, the i-th operation, the mean of the operation, and the standard deviation of the operation, respectively. i Let y be the quality data of the i-th product, and S be the average quality data of the product. y This represents the standard deviation of product quality data.

[0021] Preferably, the expression of the reactor big data model is the following expression (2):

[0022] [Z]=Φ(X,Y) (2)

[0023] Where Φ is the formal equation of the big data model, X is the input variable, including the above, Y is the hyperparameter of the big data model, and Z includes the purity of silane at the fluidized bed outlet.

[0024] Preferably, replacing the fluidized bed reactor module in the initial mechanism model with the reactor big data model to obtain the composite model of cold hydrogenation process parameters specifically includes:

[0025] The output parameters of the reactor big data model are used as the input parameters of the downstream mechanism model.

[0026] Secondly, embodiments of the present invention also provide a method for optimizing cold hydrogenation process parameters, the method comprising:

[0027] S1. Collect the initial process parameters of each piece of equipment in the polysilicon production system during the production process;

[0028] S2. Based on the initial process parameters, the predicted process parameters are obtained using the cold hydrogenation process parameter prediction method according to any one of claims 1 to 5.

[0029] S3. If the predicted process parameters do not meet the optimization objective, adjust the initial process parameters and update the initial process parameters, then return to step S2.

[0030] The initial process parameters and the predicted process parameters are output until the predicted process parameters meet the optimization objective.

[0031] S4. Adjust the actual process parameters of each piece of equipment in the polysilicon production system during production based on the initial process parameters. Thirdly, embodiments of the present invention also provide a cold hydrogenation process parameter prediction device, the device comprising:

[0032] The device includes:

[0033] The first acquisition module is used to acquire the initial process parameters of the cold hydrogenation process. The initial process parameters are the process conditions and operating parameters that are preset at the beginning of the cold hydrogenation process and used to simulate and predict the performance of the process.

[0034] The second acquisition module is used to acquire a pre-created composite model of cold hydrogenation process parameters, which includes a front-end mechanism model, a reactor big data model, and a back-end mechanism model.

[0035] The prediction module, connected to the first acquisition module and the second acquisition module respectively, is used to simulate the entire cold hydrogenation process based on the initial process parameters and using a composite model of cold hydrogenation process parameters to obtain the predicted parameters of the cold hydrogenation process.

[0036] Fourthly, embodiments of the present invention also provide a cold hydrogenation process parameter optimization device, the device comprising:

[0037] The acquisition module is used to acquire the initial process parameters of each piece of equipment in the polysilicon production system during the production process; the cold hydrogenation process parameter prediction device mentioned in the third aspect is connected to the acquisition module and is used to obtain the predicted process parameters based on the initial process parameters.

[0038] The control module, connected to both the setting module and the cold hydrogenation process parameter prediction device, is used to adjust and update the initial process parameters when the predicted process parameters do not meet the optimization target.

[0039] The cold hydrogenation process parameter prediction device is controlled to perform operations based on the initial process parameters to obtain predicted process parameters.

[0040] The initial process parameters are output once the predicted process parameters meet the optimization objective.

[0041] An optimization module is used to adjust the actual process parameters of each piece of equipment in the polysilicon production system during production based on the initial process parameters. Fifthly, embodiments of this application provide an electronic device, the device including: a processor and a memory storing computer program instructions;

[0042] When the processor executes computer program instructions, it implements the above-mentioned method for predicting or optimizing cold hydrogenation process parameters.

[0043] The cold hydrogenation process parameter prediction method of this application can obtain initial process parameters and use a composite model composed of a front-end mechanism model, a reactor big data model, and a back-end mechanism model to simulate and predict the entire process flow, thereby achieving accurate prediction and analysis of process performance. Attached Figure Description

[0044] Figure 1 : A flowchart of a method for predicting cold hydrogenation process parameters provided in an embodiment of this application;

[0045] Figure 2 : A schematic flowchart illustrating a method for building a complete cold hydrogenation mechanism model according to an embodiment of this application;

[0046] Figure 3 : A schematic flowchart illustrating a method for building a big data model of a fluidized bed reactor, provided in an embodiment of this application;

[0047] Figure 4This is a schematic diagram illustrating the embedding and replacement of a cold hydrogenation full-process mechanism model and a big data model provided in an embodiment of this application.

[0048] Figure 5 A schematic diagram comparing the calculation results of the cold hydrogenation full-process mechanism model with the test data provided in this application embodiment:

[0049] Figure 6 : A structural diagram of a cold hydrogenation process parameter prediction device provided in an embodiment of this application;

[0050] Figure 7 : A structural diagram of a cold hydrogenation process parameter optimization device provided in an embodiment of this application;

[0051] Figure 8 : A structural diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0052] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0053] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0054] Silicon tetrachloride hydrogenation technology is the process of converting silicon tetrachloride into trichlorosilane through hydrogenation. This technology effectively solves the "bottleneck" problem restricting the development of the polysilicon industry and is also an important technology for realizing the Siemens process closed-loop and reducing production costs. As an effective technology for treating polysilicon byproducts, the mainstream hydrogenation technology is cold hydrogenation. The cold hydrogenation production process involves introducing silicon powder, catalyst, silicon tetrachloride, and hydrogen into a fluidized bed reactor, where they are converted into trichlorosilane under certain temperature and pressure. The trichlorosilane is then purified through multi-stage distillation. During its production, process parameters need to be continuously optimized to achieve higher conversion rates with lower energy consumption.

[0055] The main operating conditions affecting the conversion rate of cold hydrogenation include: fluidized bed temperature, fluidized bed outlet pressure, fluidized bed pressure difference, silicon tetrachloride / hydrogen ratio and circulating hydrogen flow rate, quench tower pressure, quench tower top temperature, and coarse separator tower pressure and top temperature. Since the conversion rate of trichlorosilane cannot be measured in real time, it is necessary to predict the yield of each component in the product based on various process parameters in order to adjust the process conditions of the entire cold hydrogenation process in a timely manner. However, because the cold hydrogenation production process is a strongly coupled multi-input, multi-output dynamic system with characteristics such as time-varying, highly nonlinear, uncertain, and hysteresis, component yield prediction models based on pure mechanism modeling or pure big data modeling suffer from poor convergence and low prediction accuracy in the cold hydrogenation production process, resulting in low reliability of the prediction results.

[0056] Research has shown that mechanistic models are difficult to accurately simulate chemical reaction processes during chemical production. This results in significant discrepancies between reactor outlet component data and actual values ​​when building a full-process mechanistic model, which in turn affects the simulation of subsequent processes.

[0057] Example 1:

[0058] Based on the above research, in order to solve the problems of existing technologies, such as Figure 1 As shown, this embodiment provides a method for predicting cold hydrogenation process parameters, specifically including the following steps S101 to S103.

[0059] S101, Obtain the initial process parameters for the cold hydrogenation process.

[0060] The initial process parameters are the process conditions and operating parameters that are pre-set at the start of the cold hydrogenation process and used to simulate and predict the process performance. These initial process parameters provide the basic data required for the simulation, determining the starting point and conditions for the model simulation. By pre-setting the initial process parameters, different process conditions can be predicted and optimized.

[0061] The initial process parameters can be the process parameters of each unit in the cold hydrogenation process that are collected in real time, or they can be set according to the characteristics of the cold hydrogenation process or experience. Specifically, they can include: reaction temperature, reaction pressure, hydrogen flow rate, catalyst dosage, bed pressure drop, silicon tetrachloride flow rate, etc., which can be set according to the actual situation, and this application does not limit them.

[0062] S102, Obtain the pre-created composite model of cold hydrogenation process parameters.

[0063] The composite model of cold hydrogenation process parameters includes a front-end mechanism model, a reactor big data model, and a back-end mechanism model.

[0064] Front-end mechanism models are used to simulate the early stages of cold hydrogenation processes. They are created based on reaction mechanisms and physicochemical principles, and typically involve mechanisms such as reaction kinetics, thermodynamics, mass transfer, and heat transfer.

[0065] The reactor big data model, based on technologies such as machine learning and data mining, can capture complex nonlinear relationships and use a large amount of historical data and statistical methods to predict the reaction behavior and performance within the reactor.

[0066] The downstream mechanism model, based on the mechanism model, considers the physical and chemical processes of the downstream process and is used to simulate the later stages of the process flow, involving processes such as product separation and purification.

[0067] In this embodiment, the front-end and back-end mechanism models provide a mechanistic understanding of the process flow; the reactor big data model makes up for the shortcomings of the mechanism model in complex reaction systems and provides the utilization of actual production data; the composite model formed by the combination of the two improves the accuracy and applicability of the simulation.

[0068] S103, based on the initial process parameters, uses a composite model of cold hydrogenation process parameters to simulate the entire cold hydrogenation process and obtain the predicted parameters of the cold hydrogenation process.

[0069] Specifically, the initial process parameters are input into the composite model, which simulates each stage, calculates the relevant process parameters, and outputs the predicted parameters for the entire process flow.

[0070] Predictive parameters can include: product yield, product purity, by-product content, energy consumption indicators, process safety parameters, etc. The prediction results can be used for process optimization and decision support, helping engineers adjust process parameters to improve production efficiency and product quality.

[0071] The cold hydrogenation process parameter prediction method provided in this application can obtain initial process parameters and use a composite model composed of a front-end mechanism model, a reactor big data model, and a back-end mechanism model to simulate and predict the entire process flow, thereby achieving accurate prediction and analysis of process performance.

[0072] Optionally, the above-mentioned method for predicting cold hydrogenation process parameters may further include the following steps:

[0073] Collect historical process parameters and experimental analysis data of the cold hydrogenation process;

[0074] Based on historical process parameters and experimental analysis data, and grounded in chemical thermodynamics, reaction mechanisms, distillation theory, and fluidized bed reaction kinetics, an initial mechanism model for the entire cold hydrogenation process is established. The initial mechanism model includes at least a heat exchanger module, a fluidized bed reactor module, a quench tower module, a cooler module, a flash tank module, a distillation tower module, a hydrogen compressor module, and a pump module.

[0075] Correlation analysis was performed on historical process parameters to identify key reactor parameters, which are those that affect the performance of fluidized bed reactors.

[0076] Based on the key reactor parameters in the historical process parameters, a big data model of the reactor is obtained by training a neural network.

[0077] By replacing the fluidized bed reactor module in the initial mechanistic model with a reactor big data model, a composite model of cold hydrogenation process parameters is obtained.

[0078] The front-end mechanism model includes a heat exchanger module, while the back-end mechanism model includes a quench tower module, a cooler module, a flash tank module, a distillation tower module, a hydrogen compressor module, and a pump module.

[0079] Here, historical process parameters can include operating parameters such as temperature, pressure, flow rate, and concentration recorded during actual production. Experimental analysis data can include data such as product quality, component ratios, and by-product content analyzed in the laboratory.

[0080] Specifically, firstly, the physical and chemical principles of the entire cold hydrogenation process can be analyzed. This involves utilizing chemical thermodynamics (to predict the thermodynamic properties of substances, phase equilibrium, etc.), reaction mechanisms (to understand the reaction pathways, steps, and rates), distillation theory (to describe the separation process of substances in a distillation column), and fluidized bed reaction kinetics (to describe the reaction behavior and transport processes within a fluidized bed reactor), combined with historical data, to create an initial mechanistic model of the entire cold hydrogenation process. This initial mechanistic model, based on physical and chemical principles, provides a complete description of the entire cold hydrogenation process.

[0081] Then, by conducting correlation analysis on historical process parameters, key reactor parameters are selected, and a big data model of the reactor is trained using a neural network to establish a nonlinear mapping relationship between key reactor parameters and performance indicators. This enables more accurate prediction of reactor performance and supplements the shortcomings of mechanistic models under complex conditions.

[0082] Finally, the fluidized bed reactor module in the initial mechanistic model was replaced with a reactor big data model, resulting in a composite model of cold hydrogenation process parameters. In complex reaction systems, mechanistic models may struggle to accurately describe the actual situation, while big data models, trained on real-world data, can capture complex nonlinearities and coupling relationships. By replacing the original sensor mechanistic model with the reactor big data model, the reaction behavior inside the reactor can be simulated more accurately. Thus, the resulting composite model combines the advantages of both mechanistic and big data models, offering guidance from physicochemical principles and data-driven precision, thereby improving the accuracy and reliability of the entire cold hydrogenation process simulation.

[0083] In this embodiment, an initial mechanistic model including modules such as heat exchangers is established by collecting and analyzing historical data of the cold hydrogenation process, and a reactor big data model is trained using a neural network. Replacing the fluidized bed reactor module in the mechanistic model with the big data model forms a composite model, successfully integrating the advantages of mechanistic theory and big data analysis. This method improves the accuracy of the entire process simulation and prediction, accurately identifies and optimizes key parameters affecting reactor performance, and enhances production efficiency and product quality. Simultaneously, by accurately predicting the process, it enhances production safety and stability.

[0084] Optionally, the above-mentioned correlation analysis of historical process parameters to screen out key reactor parameters may specifically include:

[0085] The collected historical process parameters were analyzed using the Pearson correlation coefficient algorithm to determine the key reactor parameters.

[0086] The expression for the Pearson correlation coefficient analysis algorithm is as follows (1):

[0087]

[0088] Where r represents the correlation result, indicating the correlation between the operating parameters and product quality, n represents the number of operating parameter samples, i.e., the number of data pairs, and x... i Let be the i-th operating parameter, x be the average value of the operating parameters, Sx be the standard deviation of the operating parameters, and y be the mean value of the operating parameters. i Let y be the quality data of the i-th product, and S be the average quality data of the product. y This represents the standard deviation of product quality data.

[0089] The Pearson correlation coefficient is a statistical indicator that measures the degree of linear correlation between two variables, with a value ranging from -1 to 1.

[0090] r>0: indicates a positive correlation; the larger the value, the stronger the positive correlation.

[0091] r<0: indicates a negative correlation; the smaller the value, the stronger the negative correlation.

[0092] r=0: indicates that there is no linear correlation.

[0093] Specifically, a large amount of operational parameter data was collected from historical process flows. i (such as temperature, pressure, flow rate, etc.) and corresponding product quality data y i For each operating parameter, its correlation r with the product quality data is calculated using formula (1). Operating parameters with a strong positive or negative correlation with the product quality data are selected as key parameters of the reactor. The impact of the selected key parameters on reactor performance can be verified using experimental or production data. Based on the verification results, the list of key parameters can be further adjusted and confirmed.

[0094] To determine the strength of the correlation, r can be compared with a preset value (which can be set according to actual conditions, for example, 0.7). The correlation strength is then determined according to the following rules:

[0095] Strong positive correlation: r≥0.7

[0096] Moderate positive correlation: 0.4 ≤ r < 0.7

[0097] Weak correlation or no correlation: -0.4 <r<0.4

[0098] Moderate negative correlation: -0.7 <r≤-0.4

[0099] Strong negative correlation: r ≤ -0.7

[0100] Through quantitative statistical analysis, operational parameters that have a significant impact on product quality (such as cold hydrogenation conversion rate) can be accurately identified. In subsequent big data model training, using key parameters as input can improve the accuracy and generalization ability of the model. After mastering the key parameters, process conditions can be adjusted in a targeted manner to achieve refined control of the production process.

[0101] In this embodiment, by performing Pearson correlation coefficient analysis on historical process parameters and product quality data, this method effectively screened out key parameters that significantly affect reactor performance, ensuring the scientific rigor and relevance of subsequent model building. This improves the predictive accuracy and reliability of the big data model and also helps optimize process parameters and enhance product quality, providing solid data support and theoretical basis for the optimization and production control of the cold hydrogenation process.

[0102] Optionally, the expression for the above reactor big data model is the following expression (2):

[0103] [Z]=Φ(X,Y) (2)

[0104] Wherein, Φ is the formal equation of the big data model, which is usually established by machine learning algorithms (such as neural networks, support vector machines, etc.) and is used to describe the mapping relationship between input variables and output variables.

[0105] X is the input variable, which includes key parameters affecting reactor performance. Specifically, this may include, but is not limited to: reaction temperature (T, operating temperature inside the fluidized bed reactor), reaction pressure (P, operating pressure inside the reactor), and feed concentration (C). 原料 The input variables, such as the concentration of the feedstock entering the reactor (e.g., silicon tetrachloride (SiCl4) concentration), the gas-to-solid ratio (G / S, the ratio of hydrogen flow rate to solid catalyst amount), the feed rate (F, the volume or mass flow rate of the feedstock), and catalyst characteristic parameters (e.g., specific surface area, active site density of the catalyst), collectively determine the reaction conditions within the reactor, directly affecting the yield and purity of the product.

[0106] Y represents the hyperparameters of the big data model, including the model's structural parameters and training parameters, such as the number of layers in the neural network, the number of neurons per layer, and the learning rate. Specifically, it can include: model structural parameters (such as the number of layers and the number of neurons per layer) and training parameters (such as the learning rate, regularization parameters, activation function selection, and loss function type). By adjusting these hyperparameters, the model's performance can be optimized, improving prediction accuracy and generalization ability.

[0107] Z is the output variable, which can include key performance indicators such as the purity of silane at the fluidized bed outlet. (C0) SiH4 The mass fraction or mole fraction of silane in a product is an important indicator for evaluating product quality. It may also include byproduct content (such as the concentration of byproducts like hydrogen and hydrogen chloride), reaction conversion rate (the conversion rate of raw materials, reflecting the completeness of the reaction), and product yield (the ratio of the actual amount of product obtained to the theoretically calculated value).

[0108] Traditional mechanistic models may struggle to accurately describe highly nonlinear reaction processes, while big data models can learn from historical data to capture the complex nonlinear relationships between input and output variables.

[0109] In this embodiment, by constructing a reactor big data model based on expression (2), the complex nonlinear relationship between key input process parameters and reactor performance indicators is accurately characterized. This model utilizes big data and machine learning technologies to improve the prediction accuracy of key indicators such as silane purity at the fluidized bed outlet.

[0110] Optionally, the fluidized bed reactor module in the initial mechanistic model can be replaced with a reactor big data model to obtain a composite model of cold hydrogenation process parameters, specifically including:

[0111] The output parameters of the reactor big data model are used as the input parameters of the downstream mechanism model.

[0112] Specifically, reactor big data models, based on extensive historical data and advanced modeling techniques (such as neural networks), can more accurately simulate complex reactor behavior and capture nonlinear and multivariate relationships. This combines the theoretical rigor of mechanistic models with the high-precision predictive capabilities of big data models, overcoming the limitations of single models.

[0113] The front-end mechanism model mainly includes a heat exchanger module, and may also include material pretreatment units. Output parameters may include temperature, pressure, material composition, flow rate, etc., which reflect the state of the material before it enters the reactor.

[0114] The input parameters of the fluidized bed reactor during actual production are collected and used as input to the reactor big data model, which determines the boundary conditions and initial conditions of the reactor model, thereby ensuring that the reactor big data model can be inserted into the mechanistic model.

[0115] The output parameters of the reactor big data model can include product composition, temperature, pressure, and by-product content. The output parameters generated by the reactor big data model are input into the downstream mechanism model and connected with the downstream mechanism model to make the entire model process coherent.

[0116] The downstream mechanism model can include modules such as the quench tower, cooler, flash tank, distillation column, hydrogen compressor, and pumps. Based on the output of the reactor big data model, the downstream mechanism model can further simulate processes such as product cooling, separation, purification, and recovery. This ensures end-to-end simulation of the entire cold hydrogenation process, from raw material input to final product acquisition, enabling prediction and optimization.

[0117] In this embodiment, a composite model of cold hydrogenation process parameters is formed by replacing the fluidized bed reactor module in the initial mechanistic model with a reactor big data model trained on actual data. This model uses the output parameters of the upstream mechanistic model as input to the big data model, and then provides the output parameters of the big data model to the downstream mechanistic model, achieving high-precision simulation and prediction of the entire process. This method combines the theoretical rigor of the mechanistic model with the accurate predictive power of the big data model, improving the accuracy and reliability of process parameter prediction.

[0118] Example 2:

[0119] This embodiment also provides a method for optimizing cold hydrogenation process parameters, specifically including the following steps S1 to S4.

[0120] S1. Collect the initial process parameters of each piece of equipment in the polysilicon production system during the production process;

[0121] S2. Based on the initial process parameters, the predicted process parameters are obtained using the cold hydrogenation process parameter prediction method provided in any of the above embodiments;

[0122] S3. If the predicted process parameters do not meet the optimization objective, adjust and update the initial process parameters, then return to step S2.

[0123] The process continues until the predicted process parameters meet the optimization objective, at which point the initial process parameters and the predicted process parameters are output.

[0124] S4. Adjust the actual process parameters of each piece of equipment in the polysilicon production system during the production process according to the predicted process parameters.

[0125] Specifically, the parameters to be optimized are first selected, such as reaction temperature, reaction pressure, feed ratio, flow rate and residence time, catalyst dosage and type, etc.

[0126] Predict process parameters, including but not limited to:

[0127] Product quality indicators, such as the purity and yield of silane.

[0128] Byproduct content: such as the content of hydrogen, chlorides, etc.

[0129] Energy consumption and efficiency: Energy consumption per unit of product.

[0130] Equipment operating parameters: such as temperature distribution and pressure loss of the reactor.

[0131] The initial values ​​of the aforementioned initial process parameters are set based on empirical values, literature references, and considerations such as safety and equipment limitations. Optimization objectives are set according to production needs (e.g., maximizing product purity, increasing yield, reducing energy consumption, minimizing byproduct generation, improving economic efficiency, etc.). Then, production parameters of each piece of equipment in actual polysilicon production are collected as initial process parameters. These initial process parameters are input into the composite model of the cold hydrogenation process parameters obtained in the above embodiments. The model is run to obtain the corresponding predicted process parameters. The predicted process parameters are compared with the optimization objectives, and the initial process parameters are iteratively predicted and adjusted until the predetermined optimization objectives are met.

[0132] The parameters can be adjusted using the following methods:

[0133] (1) Determine the direction of adjustment. Identify the parameters that have the greatest impact on the optimization objective and determine which parameter needs to be increased or decreased.

[0134] (2) Adjustment methods: Manually adjust parameters based on experience and prediction results. Alternatively, algorithms can be used for adjustment. For example: gradient descent, which adjusts parameters in steps along the optimization direction; genetic algorithms, which simulate natural selection and genetic mutation to search for the global optimum; and particle swarm optimization, which finds the optimal parameter combination through group cooperation.

[0135] The cold hydrogenation process parameter optimization method provided in this embodiment obtains the process parameters from actual production as initial process parameters. Using the composite model and process parameter prediction method for cold hydrogenation process parameters obtained in the above embodiments, the initial process parameters are iteratively predicted and adjusted until the predetermined optimization objective is met. This method effectively combines the advantages of theoretical models and data-driven models, achieving precise optimization of key process parameters, improving product quality and yield, reducing production energy consumption and costs, and enhancing the stability and controllability of the process.

[0136] To facilitate understanding of the cold hydrogenation process parameter prediction method provided in this embodiment, a practical application description of the above-mentioned cold hydrogenation process parameter prediction method is provided here.

[0137] A method for intelligent prediction of key process parameters in cold hydrogenation using a composite model, comprising the following steps:

[0138] (i) Creating a mechanistic model of the entire cold hydrogenation process (i.e., the initial mechanistic model mentioned above).

[0139] like Figure 2 As shown, creating a mechanistic model for the entire cold hydrogenation process includes the following steps:

[0140] (1) Collect process parameter data that affect the conversion rate of the cold hydrogenation unit.

[0141] The process parameters affecting the conversion rate of the cold hydrogenation unit were screened, including: silicon tetrachloride vaporizer temperature, fluidized bed reactor temperature, reactor pressure, circulating hydrogen flow rate, silicon tetrachloride feed rate, temperatures of each heat exchanger in the heat recovery unit, quench tower top pressure, quench tower top temperature, and quench tower reflux flow rate. Table 1 below shows an example of the process parameters read.

[0142]

[0143]

[0144] Table 1

[0145] (2) Collect key test and analysis data of the cold hydrogenation production process.

[0146] Laboratory analysis data were collected from the silicon tetrachloride buffer tank, the reflux tank of the quench tower, the trichlorosilane buffer tank, and the top and side samples from the coarse fractionation tower. Table 2 below shows an example of the laboratory analysis data.

[0147]

[0148] The collected cold hydrogenation process parameter data were processed, and missing data were supplemented to retain a complete set of cold hydrogenation unit data.

[0149] The cold hydrogenation process parameter data is time series data, with each data point typically spaced 10 seconds apart. This invention supplements missing data using the mean. For example, if the data at 9:05:00 is not collected, it is considered missing data. The mean can be calculated for the five data points before 9:05:00 (data at 04:50, 04:40, 04:30, 04:20, and 04:10) and then used to fill in the missing value.

[0150] (4) The key process parameter data (i.e., the “complete set of cold hydrogenation device data” obtained in step (3) above) is used as the input variable, and the collected test analysis data is used as the output variable. The input and output variables are used as the modeling data for the mechanism model.

[0151] (5) Based on chemical thermodynamics, trichlorosilane reaction mechanism, tower distillation theory and fluidized bed reaction kinetics, a mechanism model of the entire cold hydrogenation process was developed. The reactor unit was modeled using a fixed conversion rate module to obtain the mechanism model of the entire cold hydrogenation process.

[0152] The internal reactions of the fluidized bed mainly include:

[0153] Main reaction: 3SiCl4 + Si + 2H2 = 4SiHCl3

[0154] Side reaction: Si + 3HCl = SiHCl3 + H2

[0155] The mechanistic model of the entire cold hydrogenation process mainly includes the following modules: heat exchanger, fluidized bed reactor, quench tower, cooler, flash tank, distillation column, hydrogen compressor, and pumps. The heat exchanger exchanges heat between the reactants and products; the reactor is the reaction equipment, converting trichlorosilane to silicon tetrachloride; the quench tower is a washing tower, washing unreacted silicon powder carried in the reaction products to the bottom; the cooler mainly continues to cool the reaction products, which then enter the flash tank, where hydrogen is separated from the chlorosilane; the hydrogen compressor provides hydrogen; and the distillation column separates trichlorosilane and silicon tetrachloride.

[0156] (6) Compare the simulation results with the actual test analysis data and continuously revise the mechanism model.

[0157] The simulation results are compared with the actual laboratory analysis data from the quench tower bottom, quench tower reflux tank, trichlorosilane buffer tank, and the top and side samples from the coarse fractionation tower. The chlorosilane property parameters in the mechanism model and the calculation path in the property method are continuously corrected to ensure that the deviation between the model calculation results and the actual laboratory test results is <12%. When the deviation between the model calculation results and the actual laboratory test results is <5%, the model is considered to have high accuracy.

[0158] (ii) Create a reactor big data model and embed it into the above-mentioned cold hydrogenation full-process mechanism model, and replace the reactor module in the mechanism model.

[0159] To achieve accurate prediction of the chlorosilane composition at the reactor outlet, a big data model for a cold hydrogenated fluidized bed reactor was developed, such as... Figure 3 As shown, the construction of the reactor big data model mainly includes the following steps:

[0160] (1) Conduct correlation analysis on the collected key process parameters.

[0161] Based on the key process parameters collected in step (3) of section (I) above, a correlation analysis was performed.

[0162] Correlation analysis can be performed using the Pearson correlation coefficient algorithm, as shown in the following formula (1):

[0163]

[0164] Where r represents the correlation result, n, xi, x, and Sx represent the number of sample operational parameters, the i-th operational parameter, the average value of the operational parameters, and the standard deviation of the operational parameters, respectively, and yi, y, and Sy represent the i-th product quality data, the average value of the product quality data, and the standard deviation of the product quality data, respectively.

[0165] (2) Based on the correlation analysis results, process parameters related to cold hydrogenation fluidized bed were selected for big data modeling.

[0166] Process parameters related to cold hydrogenated fluidized beds include:

[0167] Fluidized bed reactor temperature, reactor pressure, circulating hydrogen flow rate, silicon tetrachloride feed rate, silicon powder feed rate, fluidized bed pressure difference, etc.

[0168] (3) The process parameters with strong correlations are used as independent variables in the model, and the monitoring data of the cold hydrogenation fluidized bed outlet is used as the dependent variable. A big data model of the fluidized bed reactor is built using a neural network.

[0169] Using historical data of the above variables over the past year as input variables and historical data of the cold hydrogenation fluidized bed outlet monitoring indicators over the past year as output variables, the input variables and the output variables are used as modeling data for a big data model, and a big data model of the fluidized bed reactor is developed using a neural network.

[0170] (4) The formal equation of the big data model is as follows (2):

[0171] [Z]=Φ(X,Y)(2)

[0172] Where Φ is the formal equation of the big data model, X is a highly relevant process parameter, Y is the hyperparameter of the big data model, and Z is the purity of silane at the fluidized bed outlet.

[0173] (5) To avoid overfitting, the dropout of the first hidden layer is set to 0.1, that is, 10% of the neurons in this layer are randomly dropped during the training of the big data model of the cold hydrogenation reactor.

[0174] (6) The activation function of the hidden layer is the hyperbolic tangent function tanh; the activation function of the output layer is the linear function.

[0175] (7) The optimizer is selected as the learning rate adaptive algorithm adam, the loss function is the mean squared error (MSE), the model evaluation index is the mean absolute error (MAE), the weights and biases are randomly initialized, the number of iterations is 1000, and the calculation results are evaluated.

[0176] (8) Add a calibration module to the big data model of the cold hydrogenation reactor to monitor the deviation between the calculation results of the prediction model and the actual test values.

[0177] (9) The correction module compares the output (estimated value) of the prediction model with the feedback test value and decides whether to correct and update the prediction model. When there is a large difference between the two (the deviation between the predicted value and the test value is >10%. Taking the chlorosilane composition at the fluidized bed outlet as an example: test value: trichlorosilane 50%, the predicted value should be between 40% and 60%), the difference and the corresponding production conditions in the section are recorded.

[0178] (10) When such differences occur several times in a row and the corresponding production conditions are always in a stable operating state, an adjustment amount is given by calculating the deviation and superimposed on the input layer node of the prediction model. At the same time, the sample when the deviation occurs is recorded. When the amount of data accumulates to a certain amount, the prediction model is automatically retrained.

[0179] (11) The calibration module periodically calibrates the big data model of the cold hydrogenation reactor to ensure its long-term stable and reliable operation;

[0180] (12) The big data model consists of three sub-models: trichlorosilane mass ratio sub-model, dichlorosilane mass ratio sub-model, and silicon tetrachloride mass ratio sub-model, as well as a data normalization model. The three component calculation sub-models run simultaneously and output their results concurrently. The data normalization model normalizes the calculation results. The normalization formula is as follows:

[0181] X' = ​​(X - Xmin) / (Xmax - Xmin)

[0182] The output results are mainly the mass percentages of trichlorosilane, dichlorosilane, and silicon tetrachloride. The calculation results are automatically saved to the database as real-time data according to the process parameter format.

[0183] (20) Perform material balance calculations for the fluidized bed reactor, i.e., the sum of hydrogen flow rate, silicon tetrachloride feed rate and silicon powder feed rate equals the mass flow rate at the fluidized bed outlet.

[0184] (21) Collect the fluidized bed outlet temperature and pressure, and use the mass percentages of trichlorosilane, dichlorosilane, and silicon tetrachloride in step (12) and the fluidized bed outlet mass flow rate in step (20) as input variables for the boundary flow stream corresponding to the mechanistic model in the composite model, such as Figure 4 As shown.

[0185] The mechanism model is a complete cold hydrogenation process model, which includes many modules. The fluidized bed reactor is only one of these modules. Now that the fluidized bed module has been replaced by the big data module, the complete process mechanism model will be broken, which is equivalent to splitting it into two mechanism models. The first half of the fluidized bed does not need to be changed, but the second half of the mechanism model needs to input data at the boundary stream (i.e., the fluidized bed outlet). Here, the temperature, pressure, and composition of the stream are input.

[0186] (22) All parameters required for the mechanism model have been completed. The model has been debugged to ensure that the mechanism model runs normally.

[0187] (23) The front section of the composite model reactor is a mechanistic model, the reactor unit is a big data model, and the rear section of the reactor is also a mechanistic model. The physical property method of the mechanistic model of the rear section of the reactor is consistent with that of the mechanistic model of the front section.

[0188] like Figure 5 As shown in the figure, after experiments, the calculation results of the above cold hydrogenation whole process mechanism model were compared with the test data, and it can be found that the accuracy of the prediction results is high.

[0189] In this embodiment, based on the mechanism of the cold hydrogenation production process and a big data composite model, the cold hydrogenation conversion rate is predicted. According to the prediction results, abnormal production conditions are detected in advance, and production parameters are adjusted in a timely manner to achieve the goal of continuous optimization of the production system, improving the cold hydrogenation conversion rate, and enhancing the economic benefits of the unit.

[0190] Example 3:

[0191] like Figure 6 As shown, this embodiment provides a cold hydrogenation process parameter prediction device 600, including:

[0192] The first acquisition module 601 is used to acquire the initial process parameters of the cold hydrogenation process. The initial process parameters are the process conditions and operating parameters that are pre-set at the beginning of the cold hydrogenation process and used to simulate and predict the performance of the process.

[0193] The second acquisition module 602 is used to acquire a pre-created composite model of cold hydrogenation process parameters. The composite model of cold hydrogenation process parameters includes a front-end mechanism model, a reactor big data model, and a back-end mechanism model.

[0194] The prediction module 603 is connected to the first acquisition module and the second acquisition module respectively, and is used to simulate the entire cold hydrogenation process based on the initial process parameters and using a composite model of cold hydrogenation process parameters to obtain the predicted parameters of the cold hydrogenation process.

[0195] Optionally, the above-mentioned cold hydrogenation process parameter prediction device 600 includes:

[0196] The data acquisition module is used to collect historical process parameters and experimental analysis data of the cold hydrogenation process.

[0197] The first creation module, connected to the data acquisition module, is used to establish an initial mechanism model of the entire cold hydrogenation process based on historical process parameters and experimental analysis data, and on chemical thermodynamics, reaction mechanism, distillation theory and fluidized bed reaction kinetics. The initial mechanism model includes at least a heat exchanger module, a fluidized bed reactor module, a quench tower module, a cooler module, a flash tank module, a distillation tower module, a hydrogen compressor module, and a pump module.

[0198] The screening module, connected to the data acquisition module, is used to perform correlation analysis on historical process parameters and screen out key reactor parameters, which are parameters that affect the performance of the fluidized bed reactor.

[0199] The second creation module, connected to the screening module, is used to obtain a big data model of the reactor by training a neural network based on the key reactor parameters in the historical process parameters.

[0200] The third creation module, connected to both the first and second creation modules, is used to replace the fluidized bed reactor module in the initial mechanistic model with a reactor big data model, resulting in a composite model of cold hydrogenation process parameters.

[0201] The front-end mechanism model includes a heat exchanger module, while the back-end mechanism model includes a quench tower module, a cooler module, a flash tank module, a distillation tower module, a hydrogen compressor module, and a pump module.

[0202] Optionally, the above filtering module includes:

[0203] The screening unit is used to perform correlation analysis calculations on the collected historical process parameters using the Pearson correlation coefficient analysis algorithm to screen out the key parameters of the reactor.

[0204] The expression for the Pearson correlation coefficient analysis algorithm is as follows (1):

[0205]

[0206] Where r represents the correlation result, n, xi, x, and Sx represent the number of sample operations, the i-th operation, the mean of the operation, and the standard deviation of the operation, respectively, and y i Let y be the quality data of the i-th product, and S be the average quality data of the product. y This represents the standard deviation of product quality data.

[0207] In this embodiment, by obtaining initial process parameters, a composite model consisting of a front-end mechanism model, a reactor big data model, and a back-end mechanism model can be used to simulate and predict the entire process flow, thereby achieving accurate prediction and analysis of process performance.

[0208] Example 4:

[0209] like Figure 7 As shown, this embodiment provides a cold hydrogenation process parameter optimization device 700, including:

[0210] The data acquisition module 701 is used to acquire the initial process parameters of each piece of equipment in the polysilicon production system during the production process.

[0211] The cold hydrogenation process parameter prediction device 600 provided in any of the above embodiments is connected to the acquisition module and is used to obtain predicted process parameters based on initial process parameters.

[0212] The control module 702, connected to both the acquisition module 701 and the cold hydrogenation process parameter prediction device 600, is used to adjust and update the initial process parameters when the predicted process parameters do not meet the optimization target. It also controls the cold hydrogenation process parameter prediction device to execute the predicted process parameters based on the updated initial process parameters.

[0213] The initial process parameters are output once the predicted process parameters meet the optimization objective.

[0214] The optimization module 703 is connected to the control module and is used to adjust the actual process parameters of each piece of equipment in the polysilicon production system during the production process according to the initial process parameters.

[0215] Figure 8 A schematic diagram of the hardware structure of the electronic device provided in an embodiment of this application is shown.

[0216] An electronic device may include a processor 801 and a memory 802 storing computer program instructions.

[0217] Specifically, the processor 801 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0218] Memory 802 may include mass storage for data or instructions. For example, and not limitingly, memory 802 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 802 may include removable or non-removable (or fixed) media. Where appropriate, memory 802 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 802 is non-volatile solid-state memory.

[0219] Memory may include read-only memory (ROM), random access memory (RAM), disk storage media devices, optical storage media devices, flash memory devices, and electrical, optical, or other physical / tangible memory storage devices. Therefore, typically, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to one aspect of this disclosure (the above-described cold hydrogenation process parameter prediction method).

[0220] The processor 801 implements any of the scheduling methods described in the above embodiments by reading and executing computer program instructions stored in the memory 802.

[0221] In one example, the electronic device may also include a communication interface 803 and a bus 408. For example, Figure 8 As shown, the processor 801, memory 802, and communication interface 803 are connected through bus 810 and complete communication with each other.

[0222] The communication interface 803 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0223] Bus 810 includes hardware, software, or both, that couples components of an online data traffic metering device together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 810 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, any suitable bus or interconnect is contemplated herein.

[0224] Furthermore, in conjunction with the scheduling methods in the above embodiments, this application embodiment can provide a computer storage medium for implementation. The computer storage medium stores computer program instructions; when these computer program instructions are executed by a processor, they implement any of the scheduling methods in the above embodiments.

[0225] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0226] The functional blocks shown in the above block diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0227] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0228] The aspects of this disclosure have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable scheduling apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable scheduling apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by special-purpose hardware performing the specified functions or actions, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0229] The above are merely specific embodiments of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A method for predicting cold hydrogenation process parameters, characterized in that, The method includes: Obtain the initial process parameters of the cold hydrogenation process, which are the process conditions and operating parameters that are pre-set at the start of the cold hydrogenation process and used to simulate and predict the performance of the process. Obtain a pre-created composite model of cold hydrogenation process parameters, which includes a front-end mechanism model, a reactor big data model, and a back-end mechanism model; Based on the initial process parameters, the entire cold hydrogenation process is simulated using a composite model of cold hydrogenation process parameters to obtain the predicted parameters of the cold hydrogenation process.

2. The method according to claim 1, characterized in that, The method further includes: Collect historical process parameters and experimental analysis data of the cold hydrogenation process; Based on the historical process parameters and experimental analysis data, and based on chemical thermodynamics, reaction mechanism, distillation theory and fluidized bed reaction kinetics, an initial mechanism model for the entire cold hydrogenation process is established. The initial mechanism model includes at least a heat exchanger module, a fluidized bed reactor module, a quench tower module, a cooler module, a flash tank module, a distillation tower module, a hydrogen compressor module, and a pump module. Correlation analysis was performed on the historical process parameters to identify key reactor parameters, which are those that affect the performance of the fluidized bed reactor. Based on the key reactor parameters in the historical process parameters, a big data model of the reactor is obtained by training a neural network. By replacing the fluidized bed reactor module in the initial mechanistic model with the reactor big data model, the composite model of the cold hydrogenation process parameters is obtained. The front-end mechanism model includes the heat exchanger module, and the back-end mechanism model includes the quench tower module, cooler module, flash tank module, distillation tower module, hydrogen compressor module, and pump module.

3. The method according to claim 2, characterized in that, The correlation analysis of the historical process parameters to screen out key reactor parameters specifically includes: The collected historical process parameters were analyzed using the Pearson correlation coefficient algorithm to determine the key parameters of the reactor. The expression for the Pearson correlation coefficient analysis algorithm is as follows (1): Where r represents the correlation result, and n, xi, x, and S are the correlation coefficients. x y represents the number of sample operations, the i-th operation, the mean of the operation, and the standard deviation of the operation, respectively. i Let y be the quality data of the i-th product, and S be the average quality data of the product. y This represents the standard deviation of product quality data.

4. The method according to claim 2, characterized in that, The expression for the reactor big data model is as follows (2): [Z]=Φ(X,Y) (2) Where Φ is the formal equation of the big data model, X is the input variable, including the above, Y is the hyperparameter of the big data model, and Z includes the purity of silane at the fluidized bed outlet.

5. The method according to claim 2, characterized in that, The step of replacing the fluidized bed reactor module in the initial mechanism model with the reactor big data model to obtain the composite model of cold hydrogenation process parameters specifically includes: The output parameters of the reactor big data model are used as the input parameters of the downstream mechanism model.

6. A method for optimizing cold hydrogenation process parameters, characterized in that, The method includes: S1. Collect the initial process parameters of each piece of equipment in the polysilicon production system during the production process; S2. Based on the initial process parameters, the predicted process parameters are obtained using the cold hydrogenation process parameter prediction method according to any one of claims 1 to 5. S3. If the predicted process parameters do not meet the optimization objective, adjust the initial process parameters and update the initial process parameters, then return to step S2. The initial process parameters and the predicted process parameters are output until the predicted process parameters meet the optimization objective. S4. Adjust the actual process parameters of each piece of equipment in the polysilicon production system during the production process according to the initial process parameters.

7. A device for predicting cold hydrogenation process parameters, characterized in that, The device includes: The first acquisition module is used to acquire the initial process parameters of the cold hydrogenation process. The initial process parameters are the process conditions and operating parameters that are preset at the beginning of the cold hydrogenation process and used to simulate and predict the performance of the process. The second acquisition module is used to acquire a pre-created composite model of cold hydrogenation process parameters, which includes a front-end mechanism model, a reactor big data model, and a back-end mechanism model. The prediction module, connected to the first acquisition module and the second acquisition module respectively, is used to simulate the entire cold hydrogenation process based on the initial process parameters and using a composite model of cold hydrogenation process parameters to obtain the predicted parameters of the cold hydrogenation process.

8. The apparatus according to claim 7, characterized in that, The device further includes: The data acquisition module is used to collect historical process parameters and experimental analysis data of the cold hydrogenation process. The first creation module, connected to the data acquisition module, is used to establish an initial mechanism model of the entire cold hydrogenation process based on the historical process parameters and experimental analysis data, and on chemical thermodynamics, reaction mechanism, distillation theory and fluidized bed reaction kinetics. The initial mechanism model includes at least a heat exchanger module, a fluidized bed reactor module, a quench tower module, a cooler module, a flash tank module, a distillation tower module, a hydrogen compressor module, and a pump module. A screening module, connected to the data acquisition module, is used to perform correlation analysis on the historical process parameters and screen out key reactor parameters, which are parameters that affect the performance of the fluidized bed reactor. The second creation module, connected to the filtering module, is used to obtain the reactor big data model by training a neural network based on the reactor key parameters in the historical process parameters. The third creation module, connected to both the first and second creation modules, is used to replace the fluidized bed reactor module in the initial mechanism model with the reactor big data model to obtain the composite model of the cold hydrogenation process parameters. The front-end mechanism model includes the heat exchanger module, and the back-end mechanism model includes the quench tower module, cooler module, flash tank module, distillation tower module, hydrogen compressor module, and pump module.

9. The apparatus according to claim 8, characterized in that, The filtering module includes: The screening unit is used to perform correlation analysis calculations on the collected historical process parameters using the Pearson correlation coefficient analysis algorithm, and to screen out the key parameters of the reactor. The expression for the Pearson correlation coefficient analysis algorithm is as follows (1): Where r represents the correlation result, n, xi, x, and Sx represent the number of sample operations, the i-th operation, the mean of the operation, and the standard deviation of the operation, respectively, and y i Let y be the quality data of the i-th product, and S be the average quality data of the product. y This represents the standard deviation of product quality data.

10. A device for optimizing cold hydrogenation process parameters, characterized in that, The device includes: The data acquisition module is used to collect the initial process parameters of each piece of equipment in the polysilicon production system during the production process. The cold hydrogenation process parameter prediction device according to any one of claims 6 to 8 is connected to the acquisition module and is used to obtain predicted process parameters based on the set initial process parameters. The control module, connected to both the setting module and the cold hydrogenation process parameter prediction device, is used to adjust and update the initial process parameters when the predicted process parameters do not meet the optimization target. The cold hydrogenation process parameter prediction device is controlled to perform operations based on the initial process parameters to obtain predicted process parameters. The initial process parameters are output once the predicted process parameters meet the optimization objective. An optimization module, connected to the control module, is used to adjust the actual process parameters of each piece of equipment in the polysilicon production system during production based on the initial process parameters.

11. An electronic device, characterized in that, The device includes: Processor and memory storing computer program instructions; When the processor executes the computer program instructions, it implements the method as described in any one of claims 1-6.

Images