A reactor control method, device and related equipment

Abstract

The present application provides a reactor control method, apparatus and related equipment. Among them, the method includes: inputting a plurality of preset data groups into a prediction model one by one to obtain a benefit parameter corresponding to each data group in the plurality of data groups, wherein the data group corresponds to an input quantity change curve of input parameters of the reactor during a reaction process, and the benefit parameter is used to characterize the polysilicon output benefit predicted by the prediction model according to the corresponding data group; determining the data group corresponding to the highest benefit parameter as the reference data group; controlling the reactor to adjust the input parameters of the reactor according to the reference data group. By using the prediction model to determine the reference data group and controlling the reactor to adjust the input parameters of the reactor according to the reference data group, the interference of human factors can be avoided, and the polysilicon output effect of the reactor can be improved.

Description

Technical Field

[0001] This application relates to the technical field of polysilicon production, and particularly relates to a reaction furnace control method, device and related equipment. Background Art

[0002] The polysilicon reduction process is a key link in the production process. There are multiple reaction furnaces in the reduction workshop. The setting operation of the input quantity change curve of the input parameters for each reaction furnace usually refers to a pre-set optimal curve.

[0003] It is found in the application that the pre-set optimal curve is a curve obtained under continuous manual monitoring and regulation. The above-mentioned manual monitoring and regulation operations are greatly interfered by human factors. Therefore, if the reaction furnace is regulated based on the optimal curve, the actual output effect of the reaction furnace will be poor, that is, the polysilicon output effect of the reaction furnace under the relevant technical guidance is poor. Summary of the Invention

[0004] The purpose of the embodiments of this application is to provide a reaction furnace control method, device and related equipment, which are used to solve the problem of poor polysilicon output effect of the reaction furnace under the relevant technical guidance.

[0005] In a first aspect, the embodiments of this application provide a reaction furnace control method, and the method includes:

[0006] Input each of a plurality of pre-set data groups into a prediction model one by one to obtain a benefit parameter corresponding to each data group in the plurality of data groups, where the data group corresponds to the input quantity change curve of the input parameters of the reaction furnace in a reaction process, and the benefit parameter is used to characterize the polysilicon output benefit predicted by the prediction model according to the corresponding data group;

[0007] Determine the data group corresponding to the highest benefit parameter as the reference data group;

[0008] Control the reaction furnace to adjust the input parameters of the reaction furnace according to the reference data group.

[0009] Optionally, before inputting each of the plurality of pre-set data groups into the prediction model to obtain a benefit parameter corresponding to each data group in the plurality of data groups, the method further includes:

[0010] Obtain a plurality of training samples, where the training samples are used to characterize the input quantity change curve of the input parameters of the reaction furnace in a reaction process and the corresponding polysilicon output benefit in a historical period, and the curvature of any point on the input quantity change curve of the input parameters of the reaction furnace in a reaction process in the historical period is less than the corresponding curvature threshold;

[0011] Train an initial model based on the multiple training samples to obtain a prediction model.

[0012] Optionally, controlling the reactor to adjust the input parameters of the reactor according to the reference data set includes:

[0013] Determine an experimental data set of the reactor according to the data adjustment strategy corresponding to the reactor and the reference data set;

[0014] Control the reactor to adjust the input parameters of the reactor according to the experimental data set;

[0015] After controlling the reactor to adjust the input parameters of the reactor according to the reference data set, the method further includes:

[0016] Obtain the actual benefit corresponding to the experimental data set, where the actual benefit is the polysilicon output benefit of the reactor after operating based on the experimental data set;

[0017] When the actual benefit is greater than the benefit parameter corresponding to the reference data set, update the reference data set according to the experimental data set.

[0018] Optionally, the actual benefit is calculated by weighted calculation according to at least two benefit indicators corresponding to the reactor, where the at least two benefit indicators include at least two of the operation duration, polysilicon output, energy consumption, excellent product rate, trichlorosilane input amount, and hydrogen input amount corresponding to the reactor.

[0019] Optionally, obtain the operation information of the reactor, where the operation information includes the input parameters of the reactor during operation, the input amount change curve of the input parameters, temperature, voltage, current, and heat exchange power;

[0020] Statistically analyze the operation information of the reactor to generate a reactor monitoring table.

[0021] In a second aspect, an embodiment of the present application provides a reactor control device, and the device includes:

[0022] A parameter acquisition module, configured to input a preset plurality of data sets into the prediction model one by one to obtain the benefit parameter corresponding to each data set in the plurality of data sets, where the data set corresponds to the input amount change curve of the input parameters of the reactor during a reaction process, and the benefit parameter is used to characterize the polysilicon output benefit predicted by the prediction model according to the corresponding data set;

[0023] A reference determination module, configured to determine the data set corresponding to the highest benefit parameter as the reference data set;

[0024] An adjustment module, configured to control the reactor to adjust the input parameters of the reactor according to the reference data set.

[0025] Optionally, the device further includes:

[0026] A sample acquisition module, configured to acquire a plurality of training samples, where the training samples are used to characterize the input quantity change curve of the input parameters of the reactor during a reaction process in a historical period and the corresponding polysilicon production revenue, and the curvature of any point on the input quantity change curve of the input parameters of the reactor during a reaction process in the historical period is less than the corresponding curvature threshold;

[0027] A model training module, configured to train an initial model according to the plurality of training samples to obtain a prediction model.

[0028] Optionally, the adjustment module includes:

[0029] A data set adjustment sub-module, configured to determine an experimental data set of the reactor according to the data adjustment strategy corresponding to the reactor and the reference data set;

[0030] An adjustment sub-module, configured to control the reactor to adjust the input parameters of the reactor according to the experimental data set;

[0031] The device further includes:

[0032] A revenue acquisition module, configured to acquire the actual revenue corresponding to the experimental data set, where the actual revenue is the polysilicon production revenue after the reactor operates based on the experimental data set;

[0033] A reference update module, configured to update the reference data set according to the experimental data set when the actual revenue is greater than the revenue parameter corresponding to the reference data set.

[0034] In a third aspect, an embodiment of the present application further provides an electronic device, including a processor, a memory, and a computer program stored on the memory and executable on the processor. When the computer program is executed by the processor, the steps of the above-mentioned reactor control method are implemented.

[0035] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned reactor control method are implemented.

[0036] In the embodiments of the present application, by inputting multiple data groups into the prediction model one by one and comparing the multiple benefit parameters output by the prediction model, the data group corresponding to the highest benefit parameter is determined as the reference data group, and the reactor is controlled to adjust the input parameters of the reactor according to the reference data group, thereby avoiding the interference of human factors and improving the polysilicon output effect of the reactor. Description of the Drawings

[0037] Figure 1 is a flowchart of a reactor control method provided by an embodiment of the present application;

[0038] Figure 2 is a schematic structural diagram of a reactor control device provided by an embodiment of the present application;

[0039] Figure 3 is a schematic structural diagram of an electronic device provided by an embodiment of the present application. Detailed Embodiments

[0040] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are some, but not all, of the embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application.

[0041] The embodiments of the present application provide a reactor control method. Refer to Figure 1 , Figure 1 is a schematic flowchart of a reactor control method provided by an embodiment of the present application. As Figure 1 shown, the method includes the following steps:

[0042] Step 101: Input multiple preset data groups into the prediction model one by one to obtain the benefit parameter corresponding to each data group in the multiple data groups.

[0043] Among them, the data group corresponds to the input quantity change curve of the input parameters of the reactor during a reaction process, and the benefit parameter is used to characterize the polysilicon output benefit predicted by the prediction model according to the corresponding data group.

[0044] It should be noted that the multiple data groups are jointly determined according to multiple input parameters of the reactor, the initial value range of each input parameter, and the input quantity change amplitude range of each input parameter.

[0045] For example, if the multiple input parameters included in the reactor are the first input parameter, the second input parameter, and the third input parameter, where:

[0046] There are 100 optional parameter values within the initial value range of the first input parameter, and 10 optional parameter values within the input quantity variation range of the first input parameter;

[0047] There are 50 optional parameter values within the initial value range of the second input parameter, and 10 optional parameter values within the input quantity variation range of the second input parameter;

[0048] There are 100 optional parameter values within the initial value range of the third input parameter, and 10 optional parameter values within the input quantity variation range of the third input parameter;

[0049] Then 5×10 8 data groups can be generated.

[0050] The initial value range of the input parameter can be understood as the value range of the initial input quantity of the input parameter in the reaction furnace. The optional parameter values within the initial value range can be understood as the parameter values that can be selected within the initial value range based on the corresponding selection step. For example, when the initial value range is [1, 100] and the selection step is 1, there are 100 optional parameter values within the initial value range.

[0051] Similarly, the input quantity variation range of the input parameter can be understood as the input quantity increase rate or decrease rate of the input parameter in the reaction furnace.

[0052] The initial value range, selection step, and input quantity variation range of each input parameter can be determined according to actual needs, and the present application does not limit this.

[0053] Exemplarily, the above prediction model can be a model applying a recurrent neural network.

[0054] Step 102: Determine the data group corresponding to the highest revenue parameter as the reference data group.

[0055] Step 103: Control the reaction furnace to adjust the input parameters of the reaction furnace according to the reference data group.

[0056] It should be noted that controlling the reaction furnace to adjust the input parameters of the reaction furnace according to the reference data group should be understood as regulating each reaction furnace among multiple reaction furnaces so that each reaction furnace adjusts the input parameters of the reaction furnace according to the reference data group.

[0057] Among them, the installation locations of any two different reactors among the multiple reactors can be different. For example, the installation location of Reactor A is Location A, and the installation location of Reactor B is Location B. Each reactor is provided with a communication component to feedback the input quantity situation of the input parameters of the corresponding reactor to the data center, and receive the adjustment instructions issued by the data center and complete the adjustment operation of the input quantity of the input parameters of the corresponding reactor, so as to facilitate the centralized management of reactors at different locations by users and reduce the management cost of multiple reactors.

[0058] In this application, multiple data groups are input into the prediction model one by one, and multiple benefit parameters output by the prediction model are compared to determine the reference data group corresponding to the highest benefit parameter, and the reactor is controlled to adjust the input parameters of the reactor according to the reference data group, so as to avoid the interference of human factors and improve the polysilicon output effect of the reactor.

[0059] Optionally, before inputting the preset multiple data groups into the prediction model one by one to obtain the benefit parameter corresponding to each data group in the multiple data groups, the method further includes:

[0060] Obtain multiple training samples, where the training samples are used to represent the input quantity change curve of the input parameters of the reactor during a reaction process in the historical period and the corresponding polysilicon production benefit, and the curvature of any point on the input quantity change curve of the input parameters of the reactor during a reaction process in the historical period is less than the corresponding curvature threshold;

[0061] Train the initial model according to the multiple training samples to obtain a prediction model.

[0062] The input quantity change curve of the input parameters of the reactor includes two types. One type is the stable type, that is, the curvature of each point on the curve is less than the corresponding curvature threshold; the other type is the fluctuating type, indicating that there is at least one point on the curve whose curvature is greater than or equal to the corresponding curvature threshold.

[0063] The fluctuating curve indicates that there are mutations or fluctuations in the polysilicon reduction reaction in the reactor (the actual manifestation is that the input quantity of at least one input parameter in the reactor suddenly increases or decreases, and on the curve, it means that the curvature of at least one point on the curve is greater than or equal to the corresponding curvature threshold). Such mutations or fluctuations are usually abnormal reaction conditions, which will cause resource consumption and result in a lower polysilicon production benefit of the reactor during this production process (such as the occurrence of more defective products or more unqualified products).

[0064] Therefore, in the process of determining the training samples, by restricting the input quantity change curve of the determined training samples belonging to the stable class, the interference error introduced by the fluctuating class curve is avoided, so that the trained prediction model has a high prediction accuracy, and further the data accuracy of the reference data set obtained according to the prediction model is improved.

[0065] Optionally, the controlling the reactor to adjust the input parameters of the reactor according to the reference data set includes:

[0066] Determining an experimental data set of the reactor according to the data adjustment strategy corresponding to the reactor and the reference data set;

[0067] Controlling the reactor to adjust the input parameters of the reactor according to the experimental data set;

[0068] After the controlling the reactor to adjust the input parameters of the reactor according to the reference data set, the method further includes:

[0069] Obtaining the actual benefit corresponding to the experimental data set, where the actual benefit is the polysilicon output benefit of the reactor after operating based on the experimental data set;

[0070] When the actual benefit is greater than the benefit parameter corresponding to the reference data set, updating the reference data set according to the experimental data set.

[0071] It should be noted that the polysilicon reduction reaction involves hundreds of chemical reactions and many chemical reactions are reversible, which makes the simulation of the polysilicon reduction reaction difficult. Therefore, the reference data set obtained by the model prediction method cannot indicate the optimal polysilicon output benefit of the reactor.

[0072] Based on this, in order to further improve the polysilicon output benefit of the reactor, the present application adopts the method of setting a data adjustment strategy to adjust the current reference data set to complete the iterative update of the reference data set.

[0073] Among them, the data adjustment strategy can be understood as the adjustment of the initial input quantity or the change range of the input quantity of the input parameters in the reactor.

[0074] Exemplarily, if the set reference data group indicates that the initial input amount of the input parameter is 100 units, the variation range of the input amount is 1 unit / hour, the data adjustment strategy indicates that the adjustment values of the initial input amount of the input parameter include +10 units and -10 units, and the variation ranges of the input amount include +0.5 unit / hour and -0.5 unit / hour, then after adjusting the current reference data group based on the data adjustment strategy, 9 experimental data groups can be obtained, which are respectively [90 units, 0.5 unit / hour], [90 units, 1 unit / hour], [90 units, 1.5 unit / hour], [100 units, 0.5 unit / hour], [100 units, 1 unit / hour], [100 units, 1.5 unit / hour], [110 units, 0.5 unit / hour], [110 units, 1 unit / hour], [110 units, 1.5 unit / hour] in the format of [initial input amount, variation range of the input amount].

[0075] It should be noted that during the application process, if the current reference data group is updated, the testing process of subsequent experimental data groups can be cancelled to improve the testing effect and accelerate the determination efficiency of the optimal polysilicon output profit.

[0076] For example, if the multiple experimental data groups determined according to the current reference data group include the first experimental data group, the second experimental data group, and the third experimental data group, and during the process of sequentially testing the three experimental data groups, if the polysilicon output profit of the first experimental data group is greater than that of the reference data group, the first experimental data group is updated to a new reference data group, and multiple experimental data groups are determined again according to the new reference data group to continue the subsequent iterative process, that is, the second experimental data group and the third experimental data group are no longer tested.

[0077] Optionally, the actual profit is calculated by weighted calculation based on at least two profit indicators corresponding to the reaction furnace, where the at least two profit indicators include at least two of the operating duration, polysilicon output, energy consumption, high-quality product rate, trichlorosilane input amount, and hydrogen input amount corresponding to the reaction furnace.

[0078] As above, the weighted calculation method is used to improve the calculation accuracy of the actual profit; it should be noted that the calculation methods of the polysilicon output profit and the actual profit included in the training samples are the same to ensure data unity.

[0079] Among them, the shorter the operating duration of the reaction furnace, the greater the actual profit; the more the polysilicon output, the greater the actual profit; the lower the energy consumption, the greater the actual profit; the higher the high-quality product rate, the greater the actual profit; the less the trichlorosilane input amount, the greater the actual profit; the less the hydrogen input amount, the greater the actual profit.

[0080] Optionally, the method further includes:

[0081] Obtain the operation information of the reactor, where the operation information includes input parameters during the operation of the reactor, the input quantity change curve of the input parameters, temperature, voltage, current, and heat exchange power;

[0082] Statistically analyze the operation information of the reactor to generate a reactor monitoring table.

[0083] As described above, by generating a reactor monitoring table, the operation information of the reactor is visually displayed to facilitate the user's monitoring and analysis of the operation status of the reactor.

[0084] Exemplarily, the reactor monitoring table may include basic information of the reactor, control strategies, atomization data, and other measurement data;

[0085] Among them, the basic information indicates the operation batch of the reactor, operation time, communication status (indicating whether the communication component of the reactor is normally communicating with the data center), control status (in atomization problem control / non-atomization normal operation), etc.;

[0086] The control strategy indicates the set value of the hydrogen flow rate of the reactor, the set value of the trichlorosilane flow rate, etc.;

[0087] The atomization data indicates the atomization analysis measurement value of the reactor, the temperature of the tail gas in the furnace, the change amount of the tail gas temperature in the furnace, the atomization index, and the voltage change;

[0088] Other measurement data indicates the actual value of the hydrogen flow rate of the reactor, the actual value of the trichlorosilane flow rate, the actual value of the total voltage, the actual value of the infrared measurement, the molar ratio of hydrogen and trichlorosilane, etc.

[0089] In application, the reactor monitoring table may also include electrical information and heat exchange information of the reactor. Among them, the electrical information indicates the electric power, voltage, resistance, current, reference resistance, and ground current corresponding to each of the six phases of the reactor, and the heat exchange information indicates the return water volume of the heat exchange component of the reactor, the return water temperature of the heat exchange component, the water heat exchange ratio of the heat exchange component, and the water flow threshold of the heat exchange component. Among them, the six phases of the reactor indicate six groups of silicon cores arranged in the reactor (A1 phase, A2 phase, B1 phase, B2 phase, C1 phase, C2 phase), and the heat exchange component includes the furnace barrel, chassis, and jacket of the reactor.

[0090] To facilitate the user's application of the reactor monitoring table, an analysis interface may also be set up to enable the user to select some parameters in the reactor monitoring table to generate an analysis curve.

[0091] Some of the parameters in the user-selectable reactor monitoring table include the current, voltage, resistance, power, actual current rise value, current rise set value, actual hydrogen flow rate, trichlorosilane flow rate, hydrogen and trichlorosilane molar ratio, tail gas temperature, hydrogen flow set value, trichlorosilane flow set value, etc. for each of the six phases of the reactor.

[0092] See Figure 2 , Figure 2 is a structural diagram of a reactor control device 200 provided by an embodiment of the present application. As Figure 2 shown, the reactor control device 200 includes:

[0093] A parameter acquisition module 201, configured to input a plurality of preset data groups into the prediction model one by one to obtain the benefit parameter corresponding to each data group in the plurality of data groups, where the data group corresponds to the input quantity change curve of the input parameters of the reactor during a reaction process, and the benefit parameter is used to characterize the polysilicon output benefit predicted by the prediction model according to the corresponding data group;

[0094] A benchmark determination module 202, configured to determine the data group corresponding to the highest benefit parameter as the benchmark data group;

[0095] An adjustment module 203, configured to control the reactor to adjust the input parameters of the reactor according to the benchmark data group.

[0096] Optionally, the device 200 further includes:

[0097] A sample acquisition module, configured to acquire a plurality of training samples, where the training samples are used to characterize the input quantity change curve of the input parameters of the reactor during a reaction process in a historical period and the corresponding polysilicon output benefit, and the curvature of any point on the input quantity change curve of the input parameters of the reactor during a reaction process in the historical period is less than the corresponding curvature threshold;

[0098] A model training module, configured to train an initial model according to the plurality of training samples to obtain a prediction model.

[0099] Optionally, the adjustment module 203 includes:

[0100] A data group adjustment sub-module, configured to determine the experimental data group of the reactor according to the data adjustment strategy corresponding to the reactor and the benchmark data group;

[0101] An adjustment sub-module, configured to control the reactor to adjust the input parameters of the reactor according to the experimental data group;

[0102] The device 200 further includes:

[0103] A revenue acquisition module for acquiring the actual revenue corresponding to the experimental data set, where the actual revenue is the polysilicon output revenue of the reactor after operating based on the experimental data set.

[0104] A benchmark update module for updating the benchmark data set according to the experimental data set when the actual revenue is greater than the revenue parameter corresponding to the benchmark data set.

[0105] Optionally, the actual revenue is calculated by weighted calculation based on at least two revenue indicators corresponding to the reactor, where the at least two revenue indicators include at least two of the operating duration, polysilicon output, energy consumption, excellent product rate, trichlorosilane input amount, and hydrogen input amount corresponding to the reactor.

[0106] Optionally, the device 200 further includes:

[0107] An operation monitoring module for acquiring the operation information of the reactor, where the operation information includes the input parameters, the input quantity change curve of the input parameters, temperature, voltage, current, and heat exchange power during the operation of the reactor.

[0108] A statistics module for statistically analyzing the operation information of the reactor to generate a reactor monitoring table.

[0109] The reactor control device 200 provided by the embodiments of the present application can implement each process in the above method embodiments. To avoid repetition, it will not be elaborated here.

[0110] Please refer to Figure 3 , Figure 3 which is a schematic structural diagram of an electronic device provided by the embodiments of the present application. As Figure 3 shown, the electronic device may include a processor 301, a memory 302, and a program 3021 stored on the memory 302 and executable on the processor 301.

[0111] When the program 3021 is executed by the processor 301, it can implement Figure 1 any step in the corresponding method embodiment and achieve the same beneficial effects, which will not be elaborated here.

[0112] Those of ordinary skill in the art can understand that all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions, and the program can be stored in a readable medium.

[0113] The embodiments of the present application further provide a readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it can implement the above Figure 1Any step in the corresponding method embodiment, and can achieve the same technical effect. To avoid repetition, it will not be elaborated here.

[0114] The computer-readable storage medium of the embodiments of the present application can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (non-exhaustive list) of the computer-readable storage medium include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, the computer-readable storage medium can be any tangible medium that contains or stores a program, and this program can be used by or in combination with an instruction execution system, apparatus, or device.

[0115] The computer-readable signal medium can include a data signal propagated in a baseband or as part of a carrier wave, which carries the computer-readable program code. Such a propagated data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer-readable signal medium can also be any computer-readable medium other than the computer-readable storage medium, and this computer-readable medium can send, propagate, or transmit a program for use by or in combination with an instruction execution system, apparatus, or device.

[0116] The program code contained on the storage medium can be transmitted by any suitable medium, including but not limited to wireless, wire, optical fiber, RF, etc., or any suitable combination of the above.

[0117] Computer program code for performing the operations of this application can be written in one or more programming languages or combinations thereof. The programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, executed as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or terminal. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (e.g., by connecting through the Internet using an Internet service provider).

[0118] The above is the preferred implementation mode of the embodiments of this application. It should be noted that for those of ordinary skill in the art of this technology, without departing from the principle described in this application, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of this application.

Claims

1. A reactor control method, characterized in that, The method includes: Inputting a plurality of preset data groups into a prediction model one by one to obtain a benefit parameter corresponding to each data group in the plurality of data groups. Wherein, the data group corresponds to an input quantity change curve of input parameters during a reaction process of a reactor, and the benefit parameter is used to characterize the polysilicon output benefit predicted by the prediction model according to the corresponding data group. The plurality of data groups are jointly determined according to a plurality of the input parameters of the reactor, an initial value range of each input parameter, and a change amplitude range of the input quantity of each input parameter; Determining the data group corresponding to the highest benefit parameter as the reference data group; Controlling the reactor to adjust the input parameters of the reactor according to the reference data group. The controlling the reactor to adjust the input parameters of the reactor according to the reference data group includes: determining an experimental data group of the reactor according to a data adjustment strategy corresponding to the reactor and the reference data group; controlling the reactor to adjust the input parameters of the reactor according to the experimental data group.

2. The method according to claim 1, wherein Before the step of inputting a plurality of preset data groups into a prediction model one by one to obtain a benefit parameter corresponding to each data group in the plurality of data groups, the method further includes: Obtaining a plurality of training samples, wherein the training samples are used to characterize an input quantity change curve of input parameters during a reaction process of a reactor in a historical period and the corresponding polysilicon output benefit, and the curvature of any point on the input quantity change curve of the input parameters during a reaction process of the reactor in the historical period is less than the corresponding curvature threshold; Training an initial model according to the plurality of training samples to obtain a prediction model.

3. The method according to claim 1, wherein After the step of controlling the reactor to adjust the input parameters of the reactor according to the reference data group, the method further includes: Obtaining the actual benefit corresponding to the experimental data group, where the actual benefit is the polysilicon output benefit after the reactor operates based on the experimental data group; When the actual benefit is greater than the benefit parameter corresponding to the reference data group, updating the reference data group according to the experimental data group.

4. The method according to claim 3, characterized in that, The actual benefit is calculated by weighted calculation according to at least two benefit indicators corresponding to the reactor, where the at least two benefit indicators include at least two of the operation duration, polysilicon output, energy consumption, excellent product rate, trichlorosilane input quantity, and hydrogen input quantity corresponding to the reactor.

5. The method according to claim 1, characterized in that, The method further includes: Obtaining the operation information of the reactor, where the operation information includes the input parameters, the input quantity change curve of the input parameters, temperature, voltage, current, and heat exchange power during the operation of the reactor; Statistically analyzing the operation information of the reactor to generate a reactor monitoring table.

6. A reactor control device, characterized in that, The device includes: A parameter acquisition module, configured to input multiple preset data groups into a prediction model one by one, and obtain a benefit parameter corresponding to each data group in the multiple data groups, where the data group corresponds to an input quantity change curve of input parameters of a reactor during a reaction process, and the benefit parameter is used to characterize the polysilicon output benefit predicted by the prediction model according to the corresponding data group; the multiple data groups are jointly determined according to the multiple input parameters of the reactor, the initial value range of each input parameter, and the input quantity change amplitude range of each input parameter; A benchmark determination module, configured to determine the data group corresponding to the highest benefit parameter as the benchmark data group; An adjustment module, configured to control the reactor to adjust the input parameters of the reactor according to the benchmark data group. The adjustment module includes: a data group adjustment sub-module, configured to determine an experimental data group of the reactor according to a data adjustment strategy corresponding to the reactor and the benchmark data group; an adjustment sub-module, configured to control the reactor to adjust the input parameters of the reactor according to the experimental data group.

7. The device according to claim 6, characterized in that, The device further includes: A sample acquisition module, configured to acquire multiple training samples, where the training samples are used to characterize an input quantity change curve of input parameters of a reactor during a reaction process in a historical period and the corresponding polysilicon output benefit, and the curvature of any point on the input quantity change curve of the input parameters of the reactor during a reaction process in the historical period is less than a corresponding curvature threshold; A model training module, configured to train an initial model according to the multiple training samples to obtain a prediction model.

8. The device according to claim 6, characterized in that, The device further includes: A benefit acquisition module, configured to acquire the actual benefit corresponding to the experimental data group, where the actual benefit is the polysilicon output benefit after the reactor operates based on the experimental data group; A benchmark update module, configured to update the benchmark data group according to the experimental data group when the actual benefit is greater than the benefit parameter corresponding to the benchmark data group.

9. An electronic device, characterized in that, It includes a processor, a memory, and a computer program stored on the memory and executable on the processor. When the computer program is executed by the processor, the steps of the reactor control method according to any one of claims 1 to 5 are implemented.

10. A readable storage medium, characterized in that, A computer program is stored on the readable storage medium. When the computer program is executed by a processor, the steps of the reactor control method according to any one of claims 1 to 5 are implemented.

Images