Method for determining fluidized bed reaction temperature, related device and computer storage medium

Abstract

The application provides a fluidized bed reaction temperature determination method, related device and computer storage medium. The method comprises the following steps: firstly, collecting real-time data of a target factor in real time; wherein the target factor is a factor affecting the selectivity of dimethyldichlorosilane; then, pre-processing the data of the target factor to obtain first target real-time data; then, performing steady-state detection on the first target real-time data, eliminating data with large fluctuations and unstable working conditions to obtain second target real-time data; finally, inputting the second target real-time data into a fluidized bed reaction temperature determination model to obtain an optimal fluidized bed reaction temperature; wherein the fluidized bed reaction temperature determination model is obtained by training a machine learning model with at least one training sample data. Therefore, the reaction temperature control target suggestion is recommended in different production periods and working conditions, and the selectivity of dimethyldichlorosilane is the highest under the premise of ensuring yield and safety.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a method for determining the temperature of a fluidized bed reaction, related apparatus, and computer storage medium. Background Technology

[0002] Organosilicon materials are a class of high molecular polymers that are widely used in construction, daily chemicals, textile printing and dyeing, food-grade silicone rubber, food, medical and other fields, and are gradually developing towards high-end product applications, focusing on aerospace, high-end equipment, new energy, medical and health and national defense and military industries.

[0003] Methylchlorosilanes are important raw materials for manufacturing organosilicon materials, with dimethyldichlorosilane being the most in-demand. Driven by technological advancements, product innovation, and continuously increasing demand, my country's organosilicon production has maintained positive growth, making it the world's largest producer. However, overall domestic production levels remain low, with significant byproduct accumulation and low selectivity for dimethyldichlorosilane. Summary of the Invention

[0004] In view of this, this application provides a method, related apparatus and computer storage medium for determining the fluidized bed reaction temperature, which can accurately determine the optimal fluidized bed reaction temperature, thereby rapidly and effectively improving the selectivity of dimethyldichlorosilane.

[0005] The first aspect of this application provides a method for determining the temperature of a fluidized bed reaction, comprising:

[0006] Real-time data of target factors are collected; wherein, the target factors are those affecting the selectivity of dimethyldichlorosilane; the target factors include: reaction temperature, silicon powder feed rate, catalyst feed rate, fly ash rate, yield, chloromethane pressure, and bed density;

[0007] The data of the target factors are preprocessed to obtain the real-time data of the first target;

[0008] Steady-state detection is performed on the real-time data of the first target, and data with large fluctuations and unstable operating conditions are removed to obtain the real-time data of the second target;

[0009] The real-time data of the second target is input into the fluidized bed reaction temperature determination model to obtain the optimal fluidized bed reaction temperature; wherein, the fluidized bed reaction temperature determination model is obtained by training a machine learning model with at least one training sample data; the training sample data includes historical data of the target factors and the corresponding true optimal fluidized bed reaction temperatures.

[0010] Optionally, the method for constructing the fluidized bed reaction temperature determination model includes:

[0011] Obtain historical data on the target factors;

[0012] The historical data of the target factors are sorted according to the sampling time points to obtain the sorted historical data;

[0013] The sorted historical data is preprocessed to obtain the first target historical data;

[0014] The first target's historical data is transformed to obtain the second target's historical data;

[0015] Steady-state detection is performed on the historical data of the second target, and data with large fluctuations and unstable operating conditions are removed to obtain the historical data of the third target;

[0016] The historical data of the third objective is input into the machine learning model, and the output is the predicted optimal fluidized bed reaction temperature;

[0017] The machine learning model is adjusted using the error between the predicted optimal fluidized bed reaction temperature and the actual optimal fluidized bed reaction temperature until the error between the predicted optimal fluidized bed reaction temperature and the actual optimal fluidized bed reaction temperature meets a preset convergence condition. The machine learning model is then used as the fluidized bed reaction temperature determination model.

[0018] Optionally, before inputting the second target real-time data into the fluidized bed reaction temperature determination model to obtain the optimal fluidized bed reaction temperature, the method further includes:

[0019] The accuracy of the fluidized bed reaction temperature determination model was tested.

[0020] If the accuracy of the fluidized bed reaction temperature determination model does not meet the preset requirements, an online optimization model is used to optimize the fluidized bed reaction temperature determination model; wherein, the online optimization model includes: a caching and historical playback mechanism, an optimization model, and an evaluation mechanism.

[0021] Optionally, the caching and historical playback mechanism stores historical data of various relevant factors within a first preset time period after data preprocessing, optimization results, and evaluation of the optimization results.

[0022] Optionally, the optimization model is a multi-layer long short-term neural network model. The input is historical data from the previous caching and historical playback mechanism, and the model output is the optimization result for the next step.

[0023] Optionally, the evaluation mechanism evaluates the results after a second preset time following the execution of the optimization results.

[0024] Optionally, the method for determining the fluidized bed reaction temperature further includes:

[0025] The rationality of the optimization result is determined based on the optimization step size.

[0026] A second aspect of this application provides an apparatus for determining the temperature of a fluidized bed reaction, comprising:

[0027] The data acquisition unit is used to acquire real-time data of target factors; wherein, the target factors are those that affect the selectivity of dimethyldichlorosilane; the target factors include: reaction temperature, silicon powder feed rate, catalyst feed rate, fly ash rate, yield, chloromethane pressure, and bed density;

[0028] The first preprocessing unit is used to preprocess the data of the target factors to obtain the real-time data of the first target.

[0029] The first steady-state detection unit is used to perform steady-state detection on the real-time data of the first target, and after removing data with large fluctuations and unstable operating conditions, obtain the real-time data of the second target.

[0030] The first input unit is used to input the real-time data of the second target into the fluidized bed reaction temperature determination model to obtain the optimal fluidized bed reaction temperature; wherein, the fluidized bed reaction temperature determination model is obtained by training a machine learning model with at least one training sample data; the training sample data includes historical data of the target factors and the corresponding true optimal fluidized bed reaction temperature.

[0031] Optionally, the building blocks of the fluidized bed reaction temperature determination model include:

[0032] The acquisition unit is used to acquire historical data of the target factor.

[0033] The sorting unit is used to sort the historical data of the target factor according to the sampling time point to obtain the sorted historical data;

[0034] The second preprocessing unit is used to preprocess the sorted historical data to obtain the first target historical data;

[0035] A feature transformation unit is used to perform feature transformation on the first target historical data to obtain the second target historical data.

[0036] The elimination unit is used to perform steady-state detection on the historical data of the second target, and after eliminating the data with large fluctuations and unstable operating conditions, the historical data of the third target is obtained.

[0037] The second input unit is used to input the historical data of the third target into the machine learning model and output the predicted optimal fluidized bed reaction temperature.

[0038] An adjustment unit is used to adjust the machine learning model using the error between the predicted optimal fluidized bed reaction temperature and the actual optimal fluidized bed reaction temperature, until the error between the predicted optimal fluidized bed reaction temperature and the actual optimal fluidized bed reaction temperature meets a preset convergence condition, and the machine learning model is used as the fluidized bed reaction temperature determination model.

[0039] Optionally, the device for determining the fluidized bed reaction temperature further includes:

[0040] The accuracy detection unit is used to detect the accuracy of the fluidized bed reaction temperature determination model;

[0041] An online optimization unit is used to optimize the fluidized bed reaction temperature determination model if the accuracy of the model does not meet the preset requirements; wherein the online optimization model includes: a caching and historical playback mechanism, an optimization model, and an evaluation mechanism.

[0042] Optionally, the caching and historical playback mechanism stores historical data of various relevant factors within a first preset time period after data preprocessing, optimization results, and evaluation of the optimization results.

[0043] Optionally, the optimization model is a multi-layer long short-term neural network model. The input is historical data from the previous caching and historical playback mechanism, and the model output is the optimization result for the next step.

[0044] Optionally, the evaluation mechanism evaluates the results after a second preset time following the execution of the optimization results.

[0045] Optionally, the device for determining the fluidized bed reaction temperature further includes:

[0046] The rationality determination unit is used to determine the rationality of the optimization result based on the optimization step size.

[0047] A third aspect of this application provides an electronic device, comprising:

[0048] One or more processors;

[0049] A storage device on which one or more programs are stored;

[0050] When the one or more programs are executed by the one or more processors, the one or more processors implement the method for determining the fluidized bed reaction temperature as described in any one of the first aspects.

[0051] The fourth aspect of this application provides a storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method for determining the fluidized bed reaction temperature as described in any one of the first aspects.

[0052] As can be seen from the above scheme, this application provides a method, related apparatus, and computer storage medium for determining the fluidized bed reaction temperature. The method for determining the fluidized bed reaction temperature includes: first, real-time acquisition of data on target factors; wherein, the target factors are factors affecting the selectivity of dimethyldichlorosilane; the target factors include: reaction temperature, silicon powder feed rate, catalyst feed rate, fly ash rate, yield, chloromethane pressure, and bed density; then, preprocessing the data of the target factors to obtain first target real-time data; subsequently, performing steady-state detection on the first target real-time data, and removing data with large fluctuations and unstable operating conditions to obtain second target real-time data; finally, inputting the second target real-time data into the fluidized bed reaction temperature determination model to obtain the optimal fluidized bed reaction temperature; wherein, the fluidized bed reaction temperature determination model is obtained by training a machine learning model with at least one training sample data; the training sample data includes historical data of the target factors and the corresponding true optimal fluidized bed reaction temperatures. This allows for the recommendation of reaction temperature control targets under different production periods and operating conditions, achieving the highest selectivity for dimethyldichlorosilane while ensuring both yield and safety. Attached Figure Description

[0053] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0054] Figure 1 A detailed flowchart illustrating a method for determining the temperature of a fluidized bed reaction, provided in an embodiment of this application;

[0055] Figure 2 A detailed flowchart illustrating a method for constructing a fluidized bed reaction temperature determination model, as provided in another embodiment of this application;

[0056] Figure 3 A flowchart illustrating a method for determining the temperature of a fluidized bed reaction, provided as another embodiment of this application;

[0057] Figure 4 A schematic diagram of a device for determining the temperature of a fluidized bed reaction, provided for another embodiment of this application;

[0058] Figure 5 A schematic diagram of a building block for a fluidized bed reaction temperature determination model provided in another embodiment of this application;

[0059] Figure 6A schematic diagram of a device for determining the temperature of a fluidized bed reaction, provided for another embodiment of this application;

[0060] Figure 7 This is a schematic diagram of an electronic device for implementing a method for determining the temperature of a fluidized bed reaction, as provided in another embodiment of this application. Detailed Implementation

[0061] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0062] It should be noted that the concepts of "first," "second," etc., mentioned in this application are used only to distinguish different devices, modules, or units, and are not intended to limit the order of functions performed by these devices, modules, or units or their interdependencies. 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. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0063] This application provides a method for determining the temperature of a fluidized bed reaction, such as... Figure 1 As shown, the specific steps include:

[0064] S101. Real-time data collection of target factors.

[0065] Among them, the target factors are those that affect the selectivity of dimethyldichlorosilane; the target factors include: reaction temperature, silicon powder feed rate, catalyst feed rate, fly ash rate, yield, chloromethane pressure, and bed density.

[0066] In the implementation of this application, operational data and test data of historical and real-time production processes can be collected from real-time databases and laboratory information management system databases, but are not limited to those collected from real-time databases and laboratory information management system databases.

[0067] S102. Preprocess the data of the target factors to obtain the real-time data of the first target.

[0068] The preprocessing methods include, but are not limited to, data alignment, data splicing, and real-time handling of outliers and missing values ​​for online data, etc., which are not limited here.

[0069] Specifically, the real-time data of the first target is aligned according to the timestamp, and data processing such as missing value removal, outlier removal, and noise removal is performed. Common methods for handling missing values ​​include imputation and regression; common methods for outlier identification include upper and lower limits, outliers, and box plots; industrial data often contains noise, so noise processing is a necessary step, and common methods include first-order filtering and moving average filtering, which are not limited here.

[0070] S103. Perform steady-state detection on the real-time data of the first target, and remove data with large fluctuations and unstable operating conditions to obtain the real-time data of the second target.

[0071] It should be noted that, in the implementation process of this application, the real-time data of the second objective will also be saved to the real-time sample library in chronological order. The real-time sample library is a database composed of high-quality data selected after preprocessing and steady-state detection of online real-time data. The data in the real-time sample library will be closer to the current production situation than the data in the offline sample library.

[0072] Specifically, the steady-state detection method is as follows:

[0073] &abs(T F -T S )≤V T ;

[0074] Steady-state testing involves detecting continuous data within a preset time period (e.g., 2-4 hours), with fluidized bed reaction temperature being the primary monitoring metric. The specific data to be used for testing operational stability can be determined based on the specific production process.

[0075] Where, n(k) T ≤K r This indicates the temperature rate (k) during the steady-state detection period, which is 30 seconds to 5 minutes. T Less than K r The data volume, where N represents the total data volume within the steady-state detection period; V p The threshold representing the probability is determined based on the amount of data and the accuracy of the optimized model, and is typically between 70% and 95%; T F T represents the first temperature value within the steady-state detection period. S V represents the last temperature value within the steady-state detection period. T It is the threshold value for the temperature difference between the start and end points, usually based on 1-4℃.

[0076] S104. Input the real-time data of the second objective into the fluidized bed reaction temperature determination model to obtain the optimal fluidized bed reaction temperature.

[0077] The fluidized bed reaction temperature determination model is obtained by training a machine learning model with at least one training sample data; the training sample data includes historical data of the target factors and the corresponding true optimal fluidized bed reaction temperature.

[0078] Optionally, in another embodiment of this application, one implementation of the method for constructing a fluidized bed reaction temperature determination model is as follows: Figure 2 As shown, it includes:

[0079] S201. Obtain historical data of the target factors.

[0080] S202. The historical data of the target factors are sorted out according to the sampling time point to obtain the sorted historical data.

[0081] S203. Preprocess the sorted historical data to obtain the first target historical data.

[0082] Specifically, the sorted historical data is aligned according to timestamps, and processing is performed to remove missing values, outliers, and noise. Common methods for handling missing values ​​include imputation and regression; common methods for identifying outliers include upper and lower limits, outlier detection, and box plots; industrial data often contains noise, so noise processing is a necessary step, and common methods include first-order filtering and moving average filtering, which are not limited here.

[0083] S204. Perform feature transformation on the historical data of the first target to obtain the historical data of the second target.

[0084] After obtaining high-quality results through the above preprocessing, feature transformations, such as normalization and function transformation, are performed on all factors related to the selectivity of dimethyldichlorosilane. The criteria for judging the feature transformation method are the applicability and accuracy of the final optimized model.

[0085] S205. Perform steady-state testing on the historical data of the second target, and remove data with large fluctuations and unstable operating conditions to obtain the historical data of the third target.

[0086] Since optimization needs to be based on relatively stable operating conditions, it is necessary to perform steady-state detection on the historical data of the second objective and select data from the steady-state time period for subsequent model training.

[0087] It should be noted that, in the implementation process of this application, the historical data of the third target will also be saved to an offline sample library in chronological order. The offline sample library is used to store high-quality production data and corresponding timestamps selected through steady-state detection. Simultaneously, different production data will be tagged according to operating conditions, typically categorized into various abnormal operating conditions and normal production tags.

[0088] It should also be noted that the steady-state detection method can be found in step S103, and will not be repeated here.

[0089] S206. Input the historical data of the third objective into the machine learning model and output the predicted optimal fluidized bed reaction temperature.

[0090] The model for determining the fluidized bed reaction temperature can be determined based on a comprehensive evaluation of model accuracy and computation time. For machine learning models, regression models are recommended, such as backpropagation (BP) neural networks, long short-term memory (LSTM) neural networks, and support vector machines. When applying time-series regression algorithms, it's important to note that cross-validation should not be used during model parameter tuning. Training and test samples can be partitioned by time points, with the sample order arranged chronologically. For example, the training set contains data up to a certain time point, with a sample size of d%, and the sample order cannot be changed. The test set contains the remaining data, with a sample size of 1-d%, and again, the sample order cannot be changed.

[0091] S207. Determine whether the error between the predicted optimal fluidized bed reaction temperature and the actual optimal fluidized bed reaction temperature meets the preset convergence condition.

[0092] Specifically, if it is determined that the error between the predicted optimal fluidized bed reaction temperature and the actual optimal fluidized bed reaction temperature meets the preset convergence condition, then step S208 is executed; if it is determined that the error between the predicted optimal fluidized bed reaction temperature and the actual optimal fluidized bed reaction temperature does not meet the preset convergence condition, then step S209 is executed.

[0093] S208. Use machine learning models as models for determining fluidized bed reaction temperature.

[0094] S209. Adjust the machine learning model by using the error between the predicted optimal fluidized bed reaction temperature and the actual optimal fluidized bed reaction temperature.

[0095] Optionally, in another embodiment of this application, before inputting the real-time data of the second target into the fluidized bed reaction temperature determination model to obtain the optimal fluidized bed reaction temperature, one implementation of the method for determining the fluidized bed reaction temperature is as follows: Figure 3 As shown, it also includes:

[0096] S301. The accuracy of the fluidized bed reaction temperature determination model is tested.

[0097] S302. If the accuracy of the fluidized bed reaction temperature determination model does not meet the preset requirements, then the online optimization model shall be used to optimize the fluidized bed reaction temperature determination model.

[0098] The online optimization model includes: caching and historical playback mechanisms, optimization models, and evaluation mechanisms.

[0099] Optionally, in another embodiment of this application, the caching and history playback mechanism stores historical data of various relevant factors within a first preset time period after data preprocessing, optimization results, and evaluation of optimization results.

[0100] The first preset time (e.g., 1-5 hours) can be preset and changed by technicians or relevant authorized personnel, and there are no restrictions here.

[0101] Optionally, in another embodiment of this application, the optimization model is a multi-layer long short-term neural network model. The input is historical data from the previous caching and historical playback mechanism, and the model output is the optimization result for the next step.

[0102] The accuracy of the multilayer long short-time neural network model can be measured by mean squared error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE). When the accuracy of the multilayer long short-time neural network model is lower than a certain threshold, the Adam algorithm is used to optimize and update the neural network parameters.

[0103] Optionally, in another embodiment of this application, the evaluation mechanism evaluates the result after a second preset time following the execution of the optimization result.

[0104] The second preset time (e.g., 0.5-2 hours) can be preset and changed by technicians or relevant authorized personnel, and there are no restrictions here.

[0105] Specifically, a comprehensive evaluation of dimethylamine selectivity, yield, and plant stability is conducted, using the following evaluation formula:

[0106] C = a1S × P + a2 × σ T +a3×σ DT ;

[0107] Where S is the selectivity of dimethyldichlorosilane, P is the yield, and σ T To calculate the standard deviation of the reaction temperature 0.5–2 hours after the optimization results, σ DT This is the standard deviation of the reaction temperature difference 0.5-2 hours after the optimization results are implemented. The reaction temperature difference can be selected according to the actual situation, usually in the range of 30 seconds to 5 minutes. a1, a2, and a3 represent weights, and the sum of the weights is 1.

[0108] Optionally, in another embodiment of this application, one implementation of the method for determining the fluidized bed reaction temperature further includes:

[0109] The rationality of the optimization result is determined based on the optimization step size.

[0110] Specifically, the optimization step size and the magnitude of each optimization are determined through process specifications and operational experience, and the rationality of the optimization result is judged based on the optimization step size. If the optimization result exceeds a certain limit of the optimization step size (V1>(1+ε%)*V), then... S If the result is not within a reasonable range, it is considered invalid and the optimization result needs to be recalculated. If the calculation result is within a reasonable range and does not exceed the step size, the result is optimized according to the actual calculation result; if the result exceeds the step size, the result is optimized according to the optimization step size.

[0111] As can be seen from the above scheme, this application provides a method for determining the fluidized bed reaction temperature: First, real-time data of target factors are collected; wherein, the target factors are those affecting the selectivity of dimethyldichlorosilane; the target factors include: reaction temperature, silicon powder feed rate, catalyst feed rate, fly ash rate, yield, chloromethane pressure, and bed density; then, the data of the target factors are preprocessed to obtain the first target real-time data; then, steady-state detection is performed on the first target real-time data, and data with large fluctuations and unstable operating conditions are removed to obtain the second target real-time data; finally, the second target real-time data is input into the fluidized bed reaction temperature determination model to obtain the optimal fluidized bed reaction temperature; wherein, the fluidized bed reaction temperature determination model is obtained by training a machine learning model with at least one training sample data; the training sample data includes historical data of the target factors and the corresponding true optimal fluidized bed reaction temperature. Thus, it is possible to recommend reaction temperature control targets under different production periods and operating conditions, achieving the highest selectivity of dimethyldichlorosilane while ensuring yield and safety. This invention requires no changes to the original production process and equipment. It improves the selectivity of dimethylbenzene based on current production conditions, creating benefits for enterprises. It boasts advantages such as strong adaptability, high flexibility, low cost, and short cycle time. This invention is of great significance to the development of the organosilicon industry, the improvement of product quality, and the advancement of intelligent production.

[0112] Another embodiment of this application provides a device for determining the temperature of a fluidized bed reaction, such as... Figure 4 As shown, it specifically includes:

[0113] The acquisition unit 401 is used to acquire real-time data of the target factors.

[0114] Among them, the target factors are those that affect the selectivity of dimethyldichlorosilane; the target factors include: reaction temperature, silicon powder feed rate, catalyst feed rate, fly ash rate, yield, chloromethane pressure, and bed density.

[0115] The first preprocessing unit 402 is used to preprocess the data of the target factors to obtain the real-time data of the first target.

[0116] The first steady-state detection unit 403 is used to perform steady-state detection on the real-time data of the first target, and after removing data with large fluctuations and unstable operating conditions, obtain the real-time data of the second target.

[0117] The first input unit 404 is used to input the real-time data of the second target into the fluidized bed reaction temperature determination model to obtain the optimal fluidized bed reaction temperature.

[0118] The fluidized bed reaction temperature determination model is obtained by training a machine learning model with at least one training sample data; the training sample data includes historical data of the target factors and the corresponding true optimal fluidized bed reaction temperature.

[0119] For details on the specific working process of the units disclosed in the above embodiments of this application, please refer to the corresponding method embodiments, such as... Figure 1 As shown, it will not be elaborated further here.

[0120] Optionally, in another embodiment of this application, one implementation of the building blocks for the fluidized bed reaction temperature determination model is as follows: Figure 5 As shown, it includes:

[0121] Acquisition unit 501 is used to acquire historical data of the target factor.

[0122] The sorting unit 502 is used to sort the historical data of the target factor according to the sampling time point to obtain the sorted historical data.

[0123] The second preprocessing unit 503 is used to preprocess the sorted historical data to obtain the first target historical data.

[0124] The feature transformation unit 504 is used to perform feature transformation on the historical data of the first target to obtain the historical data of the second target.

[0125] The elimination unit 505 is used to perform steady-state detection on the historical data of the second target, and after eliminating the data with large fluctuations and unstable operating conditions, the historical data of the third target is obtained.

[0126] The second input unit 506 is used to input the historical data of the third objective into the machine learning model and output the predicted optimal fluidized bed reaction temperature.

[0127] The adjustment unit 507 is used to adjust the machine learning model by using the error between the predicted optimal fluidized bed reaction temperature and the actual optimal fluidized bed reaction temperature until the error between the predicted optimal fluidized bed reaction temperature and the actual optimal fluidized bed reaction temperature meets the preset convergence condition, and the machine learning model is used as the fluidized bed reaction temperature determination model.

[0128] For details on the specific working process of the units disclosed in the above embodiments of this application, please refer to the corresponding method embodiments, such as... Figure 2 As shown, it will not be elaborated further here.

[0129] Optionally, in another embodiment of this application, one implementation of the fluidized bed reaction temperature determination device is as follows: Figure 6 As shown, it also includes:

[0130] The accuracy detection unit 601 is used to detect the accuracy of the fluidized bed reaction temperature determination model.

[0131] The online optimization unit 602 is used to optimize the fluidized bed reaction temperature determination model if the accuracy of the model does not meet the preset requirements.

[0132] The online optimization model includes: caching and historical playback mechanisms, optimization models, and evaluation mechanisms.

[0133] For details on the specific working process of the units disclosed in the above embodiments of this application, please refer to the corresponding method embodiments, such as... Figure 3 As shown, it will not be elaborated further here.

[0134] Optionally, in another embodiment of this application, the caching and history playback mechanism stores historical data of various relevant factors within a first preset time period after data preprocessing, optimization results, and evaluation of optimization results.

[0135] For details on the specific working process of the units disclosed in the above embodiments of this application, please refer to the corresponding method embodiments, which will not be repeated here.

[0136] Optionally, in another embodiment of this application, the optimization model is a multi-layer long short-term neural network model. The input is historical data from the previous caching and historical playback mechanism, and the model output is the optimization result for the next step.

[0137] For details on the specific working process of the units disclosed in the above embodiments of this application, please refer to the corresponding method embodiments, which will not be repeated here.

[0138] Optionally, in another embodiment of this application, the evaluation mechanism evaluates the result after a second preset time following the execution of the optimization result.

[0139] For details on the specific working process of the units disclosed in the above embodiments of this application, please refer to the corresponding method embodiments, which will not be repeated here.

[0140] Optionally, in another embodiment of this application, one implementation of the fluidized bed reaction temperature determination device further includes:

[0141] The rationality determination unit is used to determine the rationality of the optimization result based on the optimization step size.

[0142] For details on the specific working process of the units disclosed in the above embodiments of this application, please refer to the corresponding method embodiments, which will not be repeated here.

[0143] As can be seen from the above scheme, this application provides a device for determining the fluidized bed reaction temperature: First, the acquisition unit 401 acquires real-time data of target factors; wherein, the target factors are factors affecting the selectivity of dimethyldichlorosilane; the target factors include: reaction temperature, silicon powder feed rate, catalyst feed rate, fly ash rate, yield, chloromethane pressure, and bed density; then, the first preprocessing unit 402 preprocesses the data of the target factors to obtain the first target real-time data; then, the first steady-state detection unit 403 performs steady-state detection on the first target real-time data, and removes data with large fluctuations and unstable operating conditions to obtain the second target real-time data; finally, the first input unit 404 inputs the second target real-time data into the fluidized bed reaction temperature determination model to obtain the optimal fluidized bed reaction temperature; wherein, the fluidized bed reaction temperature determination model is obtained by training a machine learning model with at least one training sample data; the training sample data includes historical data of the target factors and the corresponding true optimal fluidized bed reaction temperature. This invention allows for the recommendation of reaction temperature control targets under different production periods and operating conditions, achieving the highest selectivity for dimethyldichlorosilane while ensuring both output and safety. The invention requires no changes to the original production process and equipment, improving dimethyl selectivity based on current production conditions, thus creating benefits for enterprises. It boasts advantages such as strong adaptability, high flexibility, low cost, and short cycle time. This invention is of great significance to the development of the organosilicon industry, the improvement of product quality, and the advancement of intelligent production.

[0144] Another embodiment of this application provides an electronic device, such as... Figure 7 As shown, it includes:

[0145] One or more processors 701.

[0146] Storage device 702, on which one or more programs are stored.

[0147] When the one or more programs are executed by the one or more processors 701, the one or more processors 701 implement the method for determining the fluidized bed reaction temperature as described in any of the above embodiments.

[0148] Another embodiment of this application provides a storage medium storing a computer program thereon, wherein the computer program, when executed by a processor, implements the method for determining the fluidized bed reaction temperature as described in any of the above embodiments.

[0149] In the embodiments disclosed in this application, it should be understood that the disclosed apparatus and methods can also be implemented in other ways. The apparatus and method embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0150] Furthermore, the functional modules in the various embodiments of this disclosure can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part. If the functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this disclosure, in essence, or the part that contributes to the prior art, or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a live streaming device, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this disclosure. The aforementioned storage medium includes: USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media capable of storing program code.

[0151] Those skilled in the art will be able to implement or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for determining the temperature of a fluidized bed reaction, characterized in that, include: Real-time data of target factors are collected; wherein, the target factors are those affecting the selectivity of dimethyldichlorosilane; the target factors include: reaction temperature, silicon powder feed rate, catalyst feed rate, fly ash rate, yield, chloromethane pressure, and bed density; The data of the target factors are preprocessed to obtain the real-time data of the first target; Steady-state detection is performed on the real-time data of the first target. Data with large fluctuations and unstable operating conditions are removed to obtain the real-time data of the second target. The steady-state detection involves detecting continuous data within a preset time period, with the fluidized bed reaction temperature as the main monitoring unit. During the steady-state detection period, it is determined whether the proportion of data with a temperature rate less than a temperature rate threshold meets a preset probability threshold, and whether the temperature difference between the first and last temperature values ​​within the steady-state detection period meets a preset temperature difference threshold, so as to remove data with unstable operating conditions. The real-time data of the second target is input into the fluidized bed reaction temperature determination model to obtain the optimal fluidized bed reaction temperature. The fluidized bed reaction temperature determination model is obtained by training a machine learning model with at least one training sample data. The training sample data includes historical data of the target factors and the corresponding true optimal fluidized bed reaction temperature. The optimal fluidized bed reaction temperature is the fluidized bed reaction temperature that maximizes the selectivity of dimethyldichlorosilane under the current operating conditions.

2. The determination method according to claim 1, characterized in that, The method for constructing the fluidized bed reaction temperature determination model includes: Obtain historical data on the target factors; The historical data of the target factors are sorted according to the sampling time points to obtain the sorted historical data; The sorted historical data is preprocessed to obtain the first target historical data; The first target's historical data is transformed to obtain the second target's historical data; Steady-state detection is performed on the historical data of the second target, and data with large fluctuations and unstable operating conditions are removed to obtain the historical data of the third target; The historical data of the third objective is input into the machine learning model, and the output is the predicted optimal fluidized bed reaction temperature; The machine learning model is adjusted using the error between the predicted optimal fluidized bed reaction temperature and the actual optimal fluidized bed reaction temperature until the error between the predicted optimal fluidized bed reaction temperature and the actual optimal fluidized bed reaction temperature meets a preset convergence condition. The machine learning model is then used as the fluidized bed reaction temperature determination model.

3. The determination method according to claim 1, characterized in that, Before inputting the real-time data of the second target into the fluidized bed reaction temperature determination model to obtain the optimal fluidized bed reaction temperature, the process further includes: The accuracy of the fluidized bed reaction temperature determination model was tested. If the accuracy of the fluidized bed reaction temperature determination model does not meet the preset requirements, an online optimization model is used to optimize the fluidized bed reaction temperature determination model; wherein, the online optimization model includes: a caching and historical playback mechanism, an optimization model, and an evaluation mechanism.

4. The determination method according to claim 3, characterized in that, The caching and historical playback mechanism stores historical data of various relevant factors within a first preset time period after data preprocessing, as well as optimization results and evaluations of the optimization results.

5. The determination method according to claim 4, characterized in that, The optimization model is a multi-layer long short-term neural network model. The input is the historical data in the previous caching and historical playback mechanism, and the output is the optimization result of the next step.

6. The determination method according to claim 5, characterized in that, The evaluation mechanism evaluates the results after a second preset time following the execution of the optimization results.

7. The determination method according to claim 6, characterized in that, Also includes: The rationality of the optimization result is determined based on the optimization step size.

8. A device for determining the temperature of a fluidized bed reaction, characterized in that, include: The data acquisition unit is used to acquire real-time data of target factors; wherein, the target factors are those that affect the selectivity of dimethyldichlorosilane; the target factors include: reaction temperature, silicon powder feed rate, catalyst feed rate, fly ash rate, yield, chloromethane pressure, and bed density; The first preprocessing unit is used to preprocess the data of the target factors to obtain the real-time data of the first target. The first steady-state detection unit is used to perform steady-state detection on the real-time data of the first target, and remove data with large fluctuations and unstable operating conditions to obtain the real-time data of the second target. The steady-state detection is to detect continuous data within a preset time period, with the fluidized bed reaction temperature as the main monitoring tag. During the steady-state detection time period, it is determined whether the proportion of data with a temperature rate less than a temperature rate threshold to the total amount of data meets a preset probability threshold, and whether the temperature difference between the first temperature value and the last temperature value during the steady-state detection time period meets a preset temperature difference threshold, so as to remove data with unstable operating conditions. The first input unit is used to input the real-time data of the second target into the fluidized bed reaction temperature determination model to obtain the optimal fluidized bed reaction temperature; wherein, the fluidized bed reaction temperature determination model is obtained by training a machine learning model with at least one training sample data; the training sample data includes historical data of the target factors and the corresponding true optimal fluidized bed reaction temperature, wherein the optimal fluidized bed reaction temperature is the fluidized bed reaction temperature that enables the highest selectivity of dimethyldichlorosilane under the current operating conditions.

9. An electronic device, characterized in that, include: One or more processors; A storage device on which one or more programs are stored; When the one or more programs are executed by the one or more processors, the one or more processors implement the method for determining the fluidized bed reaction temperature as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that, It stores a computer program, wherein the computer program, when executed by a processor, implements the method for determining the fluidized bed reaction temperature as described in any one of claims 1 to 7.

Images