Intelligent temperature control method and device based on data analysis

Abstract

The present application discloses an intelligent temperature control method and device based on data analysis, the method comprising: obtaining current feeding parameters for re-feeding a quartz crucible; the quartz crucible is used to produce single crystal silicon; using a time prediction model, processing the current feeding parameters to obtain first prediction data; the first prediction data comprises: a predicted time parameter for adjusting the power of a bottom heater of the quartz crucible; wherein the time prediction model is trained using a first training sample, the first training sample uses a historical feeding parameter as an input sample, and uses a historical adjustment time corresponding to the historical feeding parameter as an output sample; at least according to the predicted time parameter, adjusting the power of the bottom heater from a first power value to a second power value to control the temperature of the silicon material in the quartz crucible, the second power value being less than the first power value.

Description

Technical Field

[0001] The present application relates to the technical field of single crystal silicon production by the Czochralski method, and in particular to an intelligent temperature control method and device based on data analysis. Background Art

[0002] At present, in the process of generating single crystal silicon by the Czochralski method, after each re-addition of materials into the quartz crucible, the solid silicon material is converted into liquid silicon material mainly by manual real-time observation of the material chemical state and manual adjustment of the heater power.

[0003] However, due to many factors such as the different feeding amount of the single crystal furnace each time, different thermal insulation, thermal field thermal insulation attenuation, etc., the lack of manual experience may lead to inaccurate power adjustment. For example, if the power is too large, the liquid temperature will be too high after all the solid silicon materials are converted into liquid silicon materials, resulting in energy waste; if the power is too low, the working time of converting solid silicon materials into liquid silicon materials will be extended.

[0004] Therefore, there is an urgent need for a technical solution that can accurately adjust the heater power. Summary of the invention

[0005] In view of the above problems, the present application provides an intelligent temperature control method and device based on data analysis to accurately adjust the heater power and achieve the purpose of accurately controlling the temperature of the silicon material in the quartz crucible. The specific scheme is as follows:

[0006] The first aspect of the present application provides an intelligent temperature control method based on data analysis, the method comprising:

[0007] Obtaining current feeding parameters for re-feeding a quartz crucible used for producing single crystal silicon;

[0008] The current feeding parameters are processed by using a time prediction model to obtain first prediction data; the first prediction data includes: a prediction time parameter for adjusting the power of the bottom heater of the quartz crucible;

[0009] The time prediction model is trained using a first training sample, wherein the first training sample uses a historical feeding parameter as an input sample and uses a historical adjustment time corresponding to the historical feeding parameter as an output sample;

[0010] At least according to the predicted time parameter, the power of the bottom heater is adjusted to switch from a first power value to a second power value to control the temperature of the silicon material in the quartz crucible, and the second power value is less than the first power value.

[0011] In a possible implementation, before adjusting the power of the bottom heater from the first power value to the second power value at least according to the predicted time parameter, the method further includes:

[0012] The current feeding parameters are processed by using a liquid-to-solid ratio prediction model to obtain second prediction data; the second prediction data includes: a predicted change rate of the liquid-to-solid ratio of the silicon material before adjusting the power of the bottom heater; the liquid-to-solid ratio prediction model is trained by using a second training sample, the second training sample uses historical feeding parameters as input samples, and uses a historical change rate of the liquid-to-solid ratio of the silicon material before the historical adjustment time as an output sample;

[0013] Wherein, adjusting the power of the bottom heater from a first power value to a second power value at least according to the predicted time parameter comprises:

[0014] Monitoring whether the current change rate of the liquid-to-solid ratio of the silicon material is consistent with the predicted change rate;

[0015] If the current change rate of the liquid-to-solid ratio of the silicon material is consistent with the predicted change rate and occurs before the start time corresponding to the predicted time parameter, at the start time corresponding to the predicted time parameter, the power of the bottom heater is adjusted to switch from the first power value to the second power value;

[0016] If the current change rate of the liquid-to-solid ratio of the silicon material is consistent with the predicted change rate within the time range corresponding to the predicted time parameter, at the moment when the current change rate of the liquid-to-solid ratio of the silicon material is consistent with the predicted change rate, adjusting the power of the bottom heater to switch from the first power value to the second power value;

[0017] If within the time range corresponding to the predicted time parameter, the current change rate of the liquid-to-solid ratio of the silicon material is not consistent with the predicted change rate, at the last moment within the time range corresponding to the predicted time parameter, the power of the bottom heater is adjusted to switch from the first power value to the second power value.

[0018] In a possible implementation, when the second power value is 0, the method further includes:

[0019] The current feeding parameters and the actual adjustment time of adjusting the power of the bottom heater are processed by using a first duration prediction model to obtain a first prediction duration; the first prediction duration represents: starting from when the power of the bottom heater is made 0 based on the actual adjustment time, the time required for the liquid-to-solid ratio of the silicon material to reach a first threshold is predicted;

[0020] The first duration prediction model is trained using a third training sample, wherein the third training sample uses the historical feeding parameters and the historical adjustment time as input samples, and uses the historical duration required for the silicon material liquid-to-solid ratio to reach a first threshold starting from when the power of the bottom heater is 0 based on the historical adjustment time as an output sample;

[0021] When the first prediction time is reached, determining whether the current liquid-to-solid ratio of the silicon material is less than the first threshold;

[0022] If the current liquid-to-solid ratio is less than the first threshold, adjusting the power of the bottom heater to a third power value; the third power value is greater than 0;

[0023] It is monitored whether a shut-down condition is satisfied, and when the shut-down condition is satisfied, the bottom heater is shut down.

[0024] In a possible implementation, the closing condition includes at least one of the following:

[0025] The duration of the bottom heater operating at the third power value reaches a duration threshold;

[0026] The bottom heater is operated at the third power value so that the current liquid-to-solid ratio increases by an increment threshold value;

[0027] The bottom heater is operated at the third power value so that the current liquid-to-solid ratio reaches the first threshold value.

[0028] In a possible implementation, when the second power value is 0, the method further includes:

[0029] The second duration prediction model is used to process the current feeding parameters and the actual adjustment time of adjusting the power of the bottom heater to obtain a second predicted duration; the second predicted duration represents: starting from when the power of the bottom heater is made 0 based on the actual adjustment time, the time required for the liquid-to-solid ratio of the silicon material to reach the second threshold is predicted;

[0030] The second duration prediction model is trained using a fourth training sample, the fourth training sample uses the historical feeding parameters and the historical adjustment time as input samples, and uses the historical duration required for the silicon material liquid-to-solid ratio to reach a second threshold starting from when the power of the bottom heater is 0 based on the historical adjustment time as an output sample;

[0031] Obtaining the actual duration of time during which the current liquid-to-solid ratio of the silicon material reaches the second threshold;

[0032] The power of the main heater of the quartz crucible is adjusted according to the actual duration and the second predicted duration.

[0033] In a possible implementation, adjusting the power of the main heater of the quartz crucible according to the actual duration and the second predicted duration includes:

[0034] Determining whether the actual duration is greater than the second predicted duration;

[0035] If the actual duration is less than the second predicted duration, controlling the power of the main heater to decrease;

[0036] If the actual duration is greater than the second predicted duration, the power of the main heater is controlled to increase.

[0037] In a possible implementation, after the current liquid-to-solid ratio reaches the second threshold, the method further includes:

[0038] Monitoring whether the current liquid-to-solid ratio of the silicon material drops below the second threshold;

[0039] If the current liquid-to-solid ratio drops below the second threshold, the power of the main heater is controlled to increase until the current liquid-to-solid ratio reaches the second threshold, and the process returns to execute: obtaining the actual duration of the current liquid-to-solid ratio of the silicon material reaching the second threshold.

[0040] In a possible implementation, after the current liquid-to-solid ratio of the silicon material in the quartz crucible reaches 100% and the quartz crucible is positioned, the method further includes:

[0041] Acquiring a current image of the silicon material in the quartz crucible;

[0042] According to the current image, obtaining the current temperature value and temperature change rate of the silicon material in the quartz crucible;

[0043] The power of the main heater of the quartz crucible is adjusted according to the current temperature value and the temperature change rate.

[0044] In a possible implementation, adjusting the power of the main heater of the quartz crucible according to the current temperature value and the temperature change rate includes:

[0045] If the temperature difference between the current temperature value and the target temperature value is greater than or equal to a first difference, and the current temperature value cannot be reduced to match the target temperature value within a target time period according to the temperature change rate, controlling the power of the main heater to be reduced;

[0046] If the temperature difference between the current temperature value and the target temperature value is greater than or equal to the first difference, and the current temperature value cannot be increased to match the target temperature value within the target time period according to the temperature change rate, the power of the main heater is controlled to increase.

[0047] A second aspect of the present application provides an intelligent temperature control device based on data analysis, the device comprising:

[0048] A parameter acquisition unit, used to obtain current feeding parameters for re-feeding a quartz crucible used for producing single crystal silicon;

[0049] A data prediction unit is used to process the current feeding parameters using a time prediction model to obtain first prediction data; the first prediction data includes: a prediction time parameter for adjusting the power of the bottom heater of the quartz crucible;

[0050] The time prediction model is trained using a first training sample, wherein the first training sample uses a historical feeding parameter as an input sample and uses a historical adjustment time corresponding to the historical feeding parameter as an output sample;

[0051] A power control unit is used to adjust the power of the bottom heater from a first power value to a second power value at least according to the predicted time parameter to control the temperature of the silicon material in the quartz crucible; the second power value is less than the first power value.

[0052] The third aspect of the present application provides a computer program product, including computer-readable instructions. When the computer-readable instructions are executed on an electronic device, the electronic device implements the intelligent temperature control method based on data analysis of the above-mentioned first aspect or any implementation method of the first aspect.

[0053] A fourth aspect of the present application provides an electronic device, comprising at least one processor and a memory connected to the processor, wherein:

[0054] The memory is used to store computer programs;

[0055] The processor is used to execute the computer program so that the electronic device can implement the intelligent temperature control method based on data analysis of the above-mentioned first aspect or any implementation method of the first aspect.

[0056] The fifth aspect of the present application provides a computer storage medium, which carries one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device can implement the intelligent temperature control method based on data analysis according to the above-mentioned first aspect or any implementation method of the first aspect.

[0057] By means of the above technical scheme, in an intelligent temperature control method and device based on data analysis provided by the present application, a training model is used to learn the relationship between the feeding parameters and the adjustment time in the historical adjustment process of the power of the bottom heater of the quartz crucible. In this way, the model obtained by training can be used to predict the time of adjusting the power according to the current feeding parameters, so as to adjust the power of the bottom heater according to the predicted time. There is no need to manually adjust the power according to experience, avoiding inaccurate adjustment caused by insufficient manual experience, thereby improving the accuracy of power adjustment, and then improving the control accuracy of the silicon material temperature in the quartz crucible. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0059] Figure 1 A schematic diagram of a flow chart of an intelligent temperature control method based on data analysis provided in an embodiment of the present application;

[0060] Figure 2 This is an example diagram of predicting time parameters in an embodiment of the present application;

[0061] Figure 3 Another schematic diagram of a flow chart of an intelligent temperature control method based on data analysis provided in an embodiment of the present application;

[0062] Figure 4 A partial flow chart of an intelligent temperature control method based on data analysis provided in an embodiment of the present application;

[0063] Figure 5 Another partial flow chart of an intelligent temperature control method based on data analysis provided in an embodiment of the present application;

[0064] Figure 6 A flowchart of another part of an intelligent temperature control method based on data analysis provided in an embodiment of the present application;

[0065] Figure 7 Another partial flow chart of an intelligent temperature control method based on data analysis provided in an embodiment of the present application;

[0066] Figure 8 A flowchart of another part of an intelligent temperature control method based on data analysis provided in an embodiment of the present application;

[0067] Fig. 9A schematic diagram of the structure of an intelligent temperature control device based on data analysis provided in an embodiment of the present application;

[0068] Fig.10 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application;

[0069] Fig.11 A flowchart of the model established in the embodiments of the present application;

[0070] Fig.12 This is a logic flow chart of intelligent temperature control in an embodiment of the present application. DETAILED DESCRIPTION

[0071] The following describes the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. The terms used in the implementation method section of the present application are only used to explain the specific embodiments of the present application, and are not intended to limit the present application.

[0072] The embodiments of the present application are described below in conjunction with the accompanying drawings. Those skilled in the art will appreciate that, with the development of technology and the emergence of new scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0073] The terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and need not be used to describe a specific order or sequential order. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances, which is only to describe the distinction mode adopted by the objects of the same attributes when describing in the embodiments of the present application. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, so that the process, method, system, product or equipment comprising a series of units need not be limited to those units, but may include other units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0074] The present application relates to the fields of photovoltaic technology and Czochralski single crystal silicon technology. The following uses the Czochralski method to produce single crystal silicon as an example to illustrate the technical solution of the present application:

[0075] In the current CZ-type silicon field, after each re-feed, the given power needs to be adjusted in real time to convert solid silicon into liquid silicon. Due to different experiences of each person, the weight of each re-feed is inconsistent, the thermal environment of each single crystal furnace is different, and the CCD (charge coupled device) camera has a limited field of view, which can easily cause the liquid temperature after temperature control to be inconsistent with the target temperature value, such as 1450℃ (degrees Celsius), resulting in extended working hours and energy consumption, affecting subsequent production. At the same time, a large number of front-line workers are required to pay attention and operate at all times, which has a certain impact on production costs, output and production efficiency.

[0076] Moreover, in the traditional single crystal growth control system, it mainly relies on manual real-time observation of the material situation and manual adjustment of the power, which has many abnormal factors such as small observation field, inability to obtain the liquid temperature when the solid material floats on the liquid, and different thickness of the solid material. And due to many factors such as different feeding amount of the single crystal furnace each time, different thermal insulation, thermal field thermal insulation attenuation, etc., a set of fixed parameters is not universal.

[0077] In view of this, in order to solve the above problems, the embodiment of the present application provides an intelligent temperature control method based on data analysis. The intelligent temperature control method based on data analysis of the embodiment of the present application is described in detail below in conjunction with the accompanying drawings.

[0078] Reference Figure 1 , Figure 1 A flow chart of an intelligent temperature control method based on data analysis provided in an embodiment of the present application, the method in this embodiment can be applied to electronic devices capable of data processing, such as computers or servers. The electronic device can control the power of the bottom heater and the main heater of the quartz crucible by sending control instructions to control the temperature of the silicon material in the quartz crucible. For example, the electronic device is a local control terminal or a cloud control terminal connected to the bottom heater and the main heater of the quartz crucible, and the control terminal sends a control instruction to the bottom heater and the main heater, and the bottom heater and the main heater heat the quartz crucible according to the power indicated by the control instruction, so that the solid silicon material in the quartz crucible is liquefied. The technical solution in this embodiment is mainly used to improve the accuracy of the heater power control, thereby improving the accuracy of the temperature control of the silicon material in the quartz crucible.

[0079] Specifically, the method in this embodiment may include the following steps:

[0080] Step 101: obtaining current feeding parameters for re-feeding a quartz crucible.

[0081] Among them, the current feeding parameters at least include: the total weight of the solid silicon material, the weight of each barrel of the re-feeding, and the re-feeding time interval between two adjacent barrels. The quartz crucible is used to produce single crystal silicon. For example, when the temperature reaches 1450°C after the silicon material in the quartz crucible is fully melted, the quartz crucible is positioned so that the quartz crucible produces single crystal silicon through the Czochralski method.

[0082] It should be noted that after the re-feeding process is completed, the current feeding parameters in the re-feeding process can be obtained in this embodiment.

[0083] Step 102: using the time prediction model, processing the current feeding parameters to obtain first prediction data.

[0084] The first prediction data includes: a predicted time parameter for adjusting the power of the bottom heater of the quartz crucible.

[0085] It should be noted that the prediction time parameters in the first prediction data may include multiple ones. Each prediction time parameter may be a time parameter including a single adjustment time, or a time range including multiple adjustment times. The adjustment time here refers to: the relative time relative to the start time of the temperature control process. The temperature control process is the process after the re-investment process. For example, the adjustment time refers to: adjusting the bottom heater at the xth minute after entering the temperature control process.

[0086] In a specific implementation, adjusting the power of the bottom heater specifically includes: adjusting the power of the bottom heater from the power value corresponding to the current power level to the power value corresponding to the next power level according to the power level of the bottom heater. The power value corresponding to the next power level here may not be 0, or may be 0 (i.e., turning off the bottom heater). In other words, adjusting the power of the bottom heater includes: reducing the power of the bottom heater to a power value that is not 0, and turning off the bottom heater (i.e., reducing the power of the bottom heater to 0).

[0087] The time prediction model can be a machine learning model, and the time prediction model can be trained using a first training sample, wherein the first training sample uses historical feeding parameters as input samples, and uses historical adjustment time corresponding to the historical feeding parameters as output samples. Historical feeding parameters refer to: the feeding parameters collected for the historical re-feeding process with manual participation. The historical adjustment time here refers to the historical moment when the power of the bottom heater is (manually) adjusted after the quartz crucible is re-fed according to the historical feeding parameters.

[0088] It should be noted that there may be multiple historical adjustment times, such as at least one historical time of reducing the power of the bottom heater and one historical time of shutting down the bottom heater. These multiple historical adjustment times may form a time list. Accordingly, in the first prediction data obtained using the time prediction model, there may also be multiple prediction time parameters, such as at least one prediction time of reducing the power of the bottom heater and one prediction time of shutting down the bottom heater. These prediction time parameters may form a time list.

[0089] In one implementation, the time prediction model includes: a first prediction model and a second prediction model.

[0090] The first prediction model can process the current feeding parameters to obtain a first prediction time for reducing the power of the bottom heater. The first prediction time can be one or more. The first prediction time represents the time for reducing the power of the bottom heater.

[0091] Specifically, the first training sample used in the first prediction model training takes the historical feeding parameters as input samples, and takes the historical time of reducing the bottom heater power corresponding to the historical feeding parameters as output samples.

[0092] The second prediction model can process the current feeding parameters to obtain a second prediction time for closing the bottom heater. The second prediction time is one. The second prediction time represents the time for closing the bottom heater.

[0093] Specifically, the first training sample used in the first prediction model training takes the historical feeding parameters as input samples, and takes the historical time of closing the bottom heater corresponding to the historical feeding parameters as output samples.

[0094] For example, Figure 2 As shown in , the predicted time parameters obtained by using the time prediction model in this embodiment include: reducing the power of the bottom heater from 80 kWh to 40 kWh at the 10th minute (the first predicted time) after entering the temperature control process, and turning off the bottom heater at the 30th minute (the second predicted time).

[0095] Step 103: adjusting the power of the bottom heater from a first power value to a second power value at least according to the predicted time parameter, so as to control the temperature of the silicon material in the quartz crucible.

[0096] The second power value is smaller than the first power value. The second power value may not be zero or may be zero.

[0097] For example, according to Figure 2 The predicted time parameters shown are: reducing the power of the bottom heater from 80 kWh to 40 kWh at the 10th minute after entering the temperature control process, turning off the bottom heater at the 30th minute, and then controlling the temperature of the silicon material in the quartz crucible by adjusting the power of the bottom heater.

[0098] It can be seen from the above scheme that in an intelligent temperature control method based on data analysis provided in an embodiment of the present application, the relationship between the feeding parameters and the adjustment time in the historical adjustment process of the power of the bottom heater of the quartz crucible is learned by training the model. In this way, the model obtained by training can be used to predict the time of adjusting the power according to the current feeding parameters, so as to adjust the power of the bottom heater according to the predicted time, and there is no need to manually adjust the power according to experience, avoiding inaccurate adjustment caused by insufficient manual experience, thereby improving the accuracy of power adjustment, and then improving the control accuracy of the silicon material temperature in the quartz crucible.

[0099] In one implementation, before step 103 in this embodiment, the following processing may also be performed, such as Figure 3 As shown in:

[0100] Step 104: using the liquid-to-solid ratio prediction model, process the current feeding parameters to obtain second prediction data.

[0101] Among them, the second prediction data includes: the predicted change rate of the liquid-to-solid ratio of the silicon material before adjusting the power of the bottom heater, that is, the predicted change rate of the liquid-to-solid ratio of the silicon material at the moment before the power of the bottom heater is adjusted. Correspondingly, each predicted change rate corresponds to a prediction time parameter. The liquid-to-solid ratio of the silicon material refers to: the proportion of liquid silicon material in all silicon materials after the silicon material is melted. For example, when the liquid-to-solid ratio of the silicon material is 0%, it indicates that the silicon material is all solid; when the liquid-to-solid ratio of the silicon material is 100%, it indicates that the silicon material is all liquid.

[0102] It should be noted that the liquid-to-solid ratio prediction model can be a machine learning model, and the liquid-to-solid ratio prediction model can be trained using the second training sample. The second training sample uses historical feeding parameters as input samples and the historical change rate of the silicon material liquid-to-solid ratio before the historical adjustment time as output samples.

[0103] There may be multiple historical adjustment moments, and accordingly, each historical adjustment moment has a corresponding historical change rate before it, and these liquid-to-solid ratio change rates form a list of change rates.

[0104] Based on this, when adjusting the power of the bottom heater in step 103, it can be achieved in the following ways: Figure 4 As shown in:

[0105] Step 401: monitor whether the current change rate of the liquid-to-solid ratio of the silicon material is consistent with the predicted change rate. If the current change rate of the liquid-to-solid ratio of the silicon material is consistent with the predicted change rate and occurs before the start time corresponding to the prediction time parameter, execute step 402; if the current change rate of the liquid-to-solid ratio of the silicon material is consistent with the predicted change rate and occurs within the time range corresponding to the prediction time parameter, execute step 403; if within the time range corresponding to the prediction time parameter, the current change rate of the liquid-to-solid ratio of the silicon material is not consistent with the predicted change rate, execute step 404.

[0106] Step 402: At the start time corresponding to the predicted time parameter, adjust the power of the bottom heater from the first power value to the second power value.

[0107] Step 403: When the current change rate of the liquid-to-solid ratio of the silicon material is consistent with the predicted change rate, adjust the power of the bottom heater from the first power value to the second power value.

[0108] Step 404: At the last moment within the time range corresponding to the predicted time parameter, adjust the power of the bottom heater from the first power value to the second power value.

[0109] Among them, the current change rate of the liquid-to-solid ratio of the silicon material can be obtained by the following method:

[0110] First, the current image of the silicon material in the quartz crucible is collected, for example, the current image of the silicon material in the quartz crucible is collected using a CCD camera; then, the current image is processed using an image segmentation model to obtain the liquid-to-solid ratio of the silicon material represented by the current image. For example, the image segmentation model can segment the image areas where the solid silicon material and the liquid silicon material are located in the current image to obtain the image area of ​​the solid silicon material and the image area of ​​the liquid silicon material, and then infer the respective proportions of the liquid silicon material and the solid silicon material according to the image area of ​​the solid silicon material and the image area of ​​the liquid silicon material, so as to obtain the liquid-to-solid ratio of the silicon material; finally, the liquid-to-solid ratio of the silicon material represented by the current image is compared with the liquid-to-solid ratio of the silicon material represented by the previous frame image to obtain the current rate of change of the liquid-to-solid ratio of the silicon material. The previous frame image here refers to the image obtained by collecting the image of the silicon material in the quartz crucible before the current image.

[0111] For example, in this embodiment, starting from entering the temperature control process, it can be determined at each moment whether the current change rate of the liquid-to-solid ratio of the silicon material is consistent with the predicted change rate corresponding to the first predicted time parameter (such as the 10th minute); if it is determined at the 9th minute that the current change rate of the liquid-to-solid ratio of the silicon material is consistent with the predicted change rate, but it has not yet reached the time range corresponding to the first predicted time parameter, that is, the starting time of the 10th minute to the 12th minute, then wait until the 10th minute to adjust the power of the bottom heater from the first power value to the second power value; if it is determined at the 11th minute that the current change rate of the liquid-to-solid ratio of the silicon material is consistent with the predicted change rate, then at the 11th minute, the power of the bottom heater is adjusted from the first power value to the second power value; if at the 12th minute, that is, the last moment, the current change rate of the liquid-to-solid ratio of the silicon material is still not consistent with the predicted change rate, then at the 12th minute, the power of the bottom heater is adjusted from the first power value to the second power value. Afterwards, at each moment, continue to determine whether the current rate of change of the liquid-to-solid ratio of the silicon material is consistent with the predicted rate of change corresponding to the second prediction time parameter (such as the 20th minute), and so on, until the last moment within the time range (28th to 30th minute) corresponding to the last prediction time parameter (such as the 30th minute), if it is determined that the current rate of change of the liquid-to-solid ratio of the silicon material is not consistent with the predicted rate of change, then the bottom heater can be turned off at the last moment (such as the 30th minute).

[0112] Therefore, in this embodiment, when adjusting the bottom heater power using the time predicted by the model, the time of the bottom heater power can be accurately adjusted within the predicted time range in combination with the current change rate of the liquid-to-solid ratio, thereby further improving the accuracy of adjusting the bottom heater power.

[0113] Based on the above implementation scheme, when the bottom heater is turned off, that is, the power of the bottom heater is adjusted to the second power value of 0, the following processing can also be performed in this embodiment, such as Figure 5 As shown in:

[0114] Step 501: using a first duration prediction model, the current feeding parameters and the actual adjustment time of adjusting the bottom heater power are processed to obtain a first predicted duration.

[0115] The first prediction duration represents: starting from when the power of the bottom heater is set to 0 based on the actual adjustment time, the time required for the liquid-to-solid ratio of the silicon material to reach the first threshold is predicted. The first threshold can be set according to demand, such as 80%. For example, the first prediction duration refers to: after the bottom heater is turned off, the time required for the liquid-to-solid ratio of the silicon material to reach 80% is predicted.

[0116] It should be noted that the actual adjustment time refers to the time for actually adjusting the bottom heater power according to the predicted time parameters, such as the “11th minute”, “28th minute”, etc., mentioned in the above examples.

[0117] The first duration prediction model can be a machine learning model, and the first duration prediction model can be trained using the third training sample. The third training sample uses historical feeding parameters and historical adjustment time as input samples, and uses the historical duration required for the silicon material liquid-to-solid ratio to reach the first threshold value starting from the power of the bottom heater being 0 based on the historical adjustment time as output samples. For example, when manual participation is made in adjusting the power of the bottom heater, it takes 10 minutes for the silicon material liquid-to-solid ratio to reach 80% after the bottom heater is turned off.

[0118] Step 502: when the first prediction time is reached, determine whether the current liquid-to-solid ratio of the silicon material is less than a first threshold value, and if so, execute step 503.

[0119] Specifically, after obtaining the first predicted duration using the first duration prediction model, as the silicon material melts, when the first predicted duration is reached, images of the silicon material in the quartz crucible are captured, and based on the captured images, the current liquid-to-solid ratio of the silicon material is obtained using the image segmentation model. If the current liquid-to-solid ratio at this time has not reached the first threshold, step 503 is executed; if the current liquid-to-solid ratio has reached the first threshold or exceeds the first threshold, no processing is performed.

[0120] Step 503: Adjust the power of the bottom heater to a third power value.

[0121] The third power value is greater than 0. For example, the third power value may be 40 kWh or 20 kWh.

[0122] For example, after the power of the bottom heater is reduced and the bottom heater is turned off, the first time prediction model is used to predict that the liquid-to-solid ratio of the silicon material may reach 80% after 20 minutes. If it is found that the current liquid-to-solid ratio of the silicon material has not reached 80% in the 20th minute after the bottom heater is turned off, it means that the residual temperature of the quartz crucible and the main heater may not be sufficient to completely melt the silicon material in time. In this case, it is necessary to reopen the bottom heater and control the bottom heater to operate at a third power value, such as 20 kWh, to continue heating the quartz crucible and promote the melting of the silicon material.

[0123] Step 504: monitor whether the closing condition is met, if yes, execute step 505, if not, continue to execute step 504.

[0124] The closing condition may include at least one of the following:

[0125] The duration of the bottom heater running at the third power value reaches a duration threshold, such as the bottom heater running at 40 kWh for 10 minutes;

[0126] The bottom heater is operated at a third power value so that the current liquid-to-solid ratio of the silicon material increases by an increment threshold, such as a 10% increase in the current liquid-to-solid ratio;

[0127] The bottom heater is operated at the third power value so that the current liquid-to-solid ratio reaches a first threshold, such as the current liquid-to-solid ratio reaches 80%.

[0128] Step 505: Turn off the bottom heater.

[0129] For example, after the bottom heater needs to be restarted and controlled to run at 40 kWh, if it lasts for 10 minutes, or the liquid-solid ratio increases by 10% or the liquid-solid ratio reaches 80%, the bottom heater can be turned off and the quartz crucible does not need to be heated by the bottom heater to avoid energy waste.

[0130] It can be seen that in this embodiment, when the residual temperature of the quartz crucible is insufficient, the bottom heater is promptly reopened, and the bottom heater is promptly closed when the closing conditions are met, thereby avoiding the situation where the bottom heater is turned on for too long and thus causing energy waste, and also avoiding the situation where the silicon material cannot be completely melted in time due to insufficient heat, resulting in extended working hours, thereby balancing energy consumption and working time consumption.

[0131] Based on the above implementation scheme, when the bottom heater is turned off, that is, the power of the bottom heater is adjusted to the second power value of 0, the following processing can also be performed in this embodiment, such as Figure 6 As shown in:

[0132] Step 601: using the second duration prediction model, the current feeding parameters and the actual adjustment time of adjusting the bottom heater power are processed to obtain a second predicted duration.

[0133] The second prediction duration represents: starting from the actual adjustment time to make the power of the bottom heater 0, the time required for the liquid-to-solid ratio of the silicon material to reach the second threshold is predicted. The second threshold is greater than the first threshold, and the second threshold can be set according to demand, such as 90%. For example, the second prediction duration refers to: after the bottom heater is turned off, the predicted time required for the liquid-to-solid ratio of the silicon material to reach 90%.

[0134] It should be noted that the second duration prediction model can be a machine learning model, and the second duration prediction model can be trained using the fourth training sample. The fourth training sample uses historical feeding parameters and historical adjustment time as input samples, and uses the historical duration required for the silicon material liquid-to-solid ratio to reach the second threshold from the power of the bottom heater being 0 based on the historical adjustment time as an output sample. For example, when manual participation is made in adjusting the power of the bottom heater, it takes 20 minutes for the silicon material liquid-to-solid ratio to reach 90% after the bottom heater is turned off.

[0135] Step 602: monitor whether the current liquid-to-solid ratio of the silicon material reaches a second threshold value. When the current liquid-to-solid ratio of the silicon material reaches the second threshold value, execute step 603; if not, return to continue to execute step 602;

[0136] Step 603: Obtain the actual duration of the current liquid-to-solid ratio of the silicon material reaching the second threshold.

[0137] Specifically, after the bottom heater is turned off, as the silicon material melts, in this embodiment, images of the silicon material in the quartz crucible are continuously captured, thereby continuously obtaining the latest current liquid-to-solid ratio of the silicon material based on the captured images using the image segmentation model. When the current liquid-to-solid ratio reaches the second threshold, the duration from the turning off of the bottom heater to this moment, that is, the actual duration, is recorded.

[0138] Step 604: adjusting the power of the main heater of the quartz crucible according to the actual duration and the second predicted duration.

[0139] In this embodiment, the power of the main heater can be increased or decreased based on the size relationship between the actual duration and the second predicted duration, so as to timely control the melting degree of the silicon material.

[0140] Specifically, step 604 can be implemented in the following manner:

[0141] First, determine whether the actual duration is greater than the second predicted duration; if the actual duration is greater than the second predicted duration, it indicates that the current power of the main heater may not be sufficient to melt the silicon material in the quartz crucible completely and in time. At this time, the power of the main heater can be controlled to increase. If the actual duration is less than the second predicted duration, it indicates that the current power of the main heater may cause the silicon material to be heated too quickly, resulting in excessive temperature in the later stage. At this time, the power of the main heater can be controlled to decrease.

[0142] It should be noted that the power of the main heater can be adjusted within a limited range when adjusting the power of the main heater in step 604. For example, the power of the main heater is controlled to increase or decrease within a limited range and not exceed the limited range. Specifically, the adjustment range of the power of the main heater can be related to the difference between the actual duration and the second predicted duration. For example, the greater the difference between the actual duration and the second predicted duration, the greater the value of the corresponding control power increase of the main heater; the smaller the difference between the actual duration and the second predicted duration, the smaller the value of the corresponding control power increase of the main heater; the greater the difference between the actual duration and the second predicted duration, the greater the value of the corresponding control power reduction of the main heater; the smaller the difference between the actual duration and the second predicted duration, the smaller the value of the corresponding control power reduction of the main heater.

[0143] Furthermore, since the liquid-to-solid ratio is realized by image acquisition, and solid silicon material may be blocked or exceed the image acquisition range within the image acquisition range, and as the liquid flows, the solid silicon material that floats out of the image acquisition range floats back into the image acquisition range, which will reduce the current liquid-to-solid ratio. Therefore, based on the above implementation scheme, after the current liquid-to-solid ratio reaches the second threshold in step 602, the method in this embodiment may further include the following processing, such as Figure 7 As shown in:

[0144] Step 605: monitor whether the current liquid-to-solid ratio of the silicon material drops below the second threshold value, if yes, execute step 606, if no, continue to execute step 605 until the liquid-to-solid ratio of the silicon material reaches 100%;

[0145] The current liquid-to-solid ratio drops below the second threshold, which means that the current liquid-to-solid ratio drops to a third threshold, and the third threshold is any value smaller than the second threshold.

[0146] Step 606: Control the power of the main heater to increase, and return to step 602 until the current liquid-to-solid ratio reaches the second threshold, and then execute step 603 to continue adjusting the power of the main heater.

[0147] The step 606 of controlling the power of the main heater to increase may specifically include setting the power of the main heater to be a multiple of the seeding power. The seeding power here refers to the power used in the seeding process.

[0148] It can be seen that in this embodiment, after the current liquid-to-solid ratio of the silicon material reaches the second threshold value, the current liquid-to-solid ratio continues to be monitored. If it is found that the current liquid-to-solid ratio falls below the second threshold value, the power of the main heater is appropriately increased to accelerate the melting of the solid silicon material. When the current liquid-to-solid ratio of the silicon material rises to the second threshold value, the relationship between the actual duration and the second predicted duration is judged again, and the power of the main heater is further increased or decreased, thereby avoiding the situation where energy waste is caused by excessive power, and the situation where the silicon material cannot be completely melted in time due to insufficient power, resulting in extended working hours, thereby balancing energy consumption and working time consumption.

[0149] In one implementation, after the current liquid-to-solid ratio of the silicon material in the quartz crucible reaches 100% and the quartz crucible is positioned, the method in this embodiment may further include the following processing, such as Figure 8 As shown in:

[0150] Step 801: Acquire the current image of silicon material in the quartz crucible.

[0151] Specifically, in this embodiment, a CCD camera may be used to collect the current image of the silicon material in the quartz crucible.

[0152] Step 802: According to the current image, obtain the current temperature value and temperature change rate of the silicon material in the quartz crucible.

[0153] Specifically, in this embodiment, the grayscale value in the current image can be analyzed, and then the current temperature value of the silicon material can be obtained according to the grayscale value, and then the current temperature value can be compared with the temperature value obtained before to obtain the latest temperature change rate.

[0154] Step 803: adjusting the power of the main heater of the quartz crucible according to the current temperature value and the temperature change rate.

[0155] In this embodiment, the temperature change rate can be combined to increase or decrease the power of the main heater according to the difference between the current temperature value and the target temperature value, so as to avoid energy waste caused by excessive power or energy waste caused by insufficient power to achieve a balance.

[0156] Specifically, in step 803, at least one of the following may be implemented in the following manner:

[0157] In one implementation, it is determined whether the temperature difference by which the current temperature value exceeds the target temperature value is greater than or equal to a first difference. If the temperature difference by which the current temperature value exceeds the target temperature value is less than the first difference, the power of the main heater may not be adjusted. If the temperature difference by which the current temperature value exceeds the target temperature value is greater than or equal to the first difference, it is determined whether the current temperature value can be reduced to match the target temperature value within the target time period according to the temperature change rate. If so, there is no need to adjust the power of the main heater. If the current temperature value cannot be reduced to match the target temperature value within the target time period according to the temperature change rate, it is necessary to adjust the power of the main heater, that is, control the power reduction of the main heater to avoid the situation where the power is too high and the heat is too much, resulting in the inability to reduce the current temperature value.

[0158] In another implementation, it is determined whether the temperature difference between the current temperature value and the target temperature value is greater than or equal to the first difference. If the temperature difference between the current temperature value and the target temperature value is less than the first difference, the power of the main heater may not be adjusted. If the temperature difference between the current temperature value and the target temperature value is greater than or equal to the first difference, it is determined whether the current temperature value can be increased to match the target temperature value within the target time period according to the temperature change rate. If so, it is not necessary to adjust the power of the main heater. If the current temperature value cannot be increased to match the target temperature value within the target time period according to the temperature change rate, it is necessary to adjust the power of the main heater, that is, control the power increase of the main heater to timely add heat to the quartz crucible so that the temperature of the liquid silicon material in the quartz crucible can match the target temperature value.

[0159] Wherein, the current temperature value matches the target temperature value, which can be: the temperature difference between the current temperature value and the target temperature value is less than the first difference. Correspondingly, whether the current temperature value can be reduced to match the target temperature value within the target time length according to the temperature change rate means: whether the current temperature value can be reduced to match the target temperature value within the target time length according to the temperature change rate is less than the first difference. And whether the current temperature value can be increased to match the target temperature value within the target time length according to the temperature change rate means: whether the current temperature value can be increased to match the target temperature value within the target time length according to the temperature change rate is less than the first difference.

[0160] The target duration here can be set according to the demand, such as 3 minutes. The first difference value can be 10° C. The target temperature value refers to the target temperature for controlling the temperature of the silicon material, such as the melting point of the silicon material 1450° C.

[0161] For example, if the current temperature of the silicon material is 1480°C, the temperature difference from 1450°C is 30°C, which is 10°C higher than the first difference, and the temperature change rate is 1°C per minute. At this time, according to the temperature change rate, the current temperature value cannot be reduced to between 1440°C and 1460°C. In this case, the power of the main heater should be controlled to reduce to avoid excessively high temperature of the silicon material. The temperature range of 1440°C to 1460°C here refers to the range where the temperature difference from 1450°C is less than 10°C.

[0162] For another example, if the current temperature of the silicon material is 1470°C and the temperature difference exceeds 1450°C by 20°C, which has exceeded the first difference of 10°C, and the temperature change rate is: decreasing by 5°C per minute, then, according to the temperature change rate, the current temperature value can be reduced to between 1440°C and 1460°C within 3 minutes. In this case, the power of the main heater does not need to be adjusted, and the current temperature value can also match the target temperature value of 1450°C within 3 minutes.

[0163] For another example, if the current temperature value of the silicon material is 1420°C, the temperature difference below 1450°C is 30°C, which has exceeded the first difference of 10°C, and the temperature change rate is: 2°C per minute, then, according to the temperature change rate, the current temperature value can only increase to 1426°C within 3 minutes. Then, at this time, the power increase of the main heater should be controlled to increase the heat in the quartz crucible in time, increase the liquid temperature of the silicon material, and avoid solidification of the silicon material.

[0164] For another example, if the current temperature value of the silicon material is 1420°C, the temperature difference below 1450°C is 30°C, which has exceeded the first difference of 10°C, and the temperature change rate is: decreasing by 1°C per minute, then, according to the temperature change rate, the current temperature value must not increase to between 1440°C and 1460°C. In this case, the power increase of the main heater must be controlled to increase the heat in the quartz crucible in time, increase the liquid temperature of the silicon material, and avoid solidification of the silicon material.

[0165] For another example, if the current temperature value of the silicon material is 1420°C, the temperature difference below 1450°C is 30°C, which has exceeded the first difference of 10°C, and the temperature change rate is: 8°C per minute, then, according to the temperature change rate, the current temperature value can be increased to 1444°C within 3 minutes. Then, at this time, the power of the main heater does not need to be adjusted, and the current temperature value can also match the target temperature value of 1450°C within 3 minutes.

[0166] The above introduces an intelligent temperature control method based on data analysis provided in an embodiment of the present application. The following will introduce a device for executing the above-mentioned intelligent temperature control method based on data analysis.

[0167] See also Fig. 9, is a structural schematic diagram of an intelligent temperature control device based on data analysis provided in an embodiment of the present application. The device in this embodiment can be configured in an electronic device capable of data processing, such as a computer or a server. The electronic device can control the power of the bottom heater and the main heater of the quartz crucible by sending control instructions to control the temperature of the silicon material in the quartz crucible. For example, the electronic device is a local control terminal or a cloud control terminal connected to the bottom heater and the main heater of the quartz crucible, and the control terminal sends a control instruction to the bottom heater and the main heater, and the bottom heater and the main heater heat the quartz crucible according to the power indicated by the control instruction to liquefy the solid silicon material in the quartz crucible. The technical solution in this embodiment is mainly used to improve the accuracy of the heater power control, thereby improving the accuracy of the temperature control of the silicon material in the quartz crucible.

[0168] Specifically, the device in this embodiment may include the following units:

[0169] The parameter acquisition unit 901 is used to obtain the current feeding parameters for re-feeding the quartz crucible;

[0170] The data prediction unit 902 is used to process the current feeding parameters using a time prediction model to obtain first prediction data; the first prediction data includes: a prediction time parameter for adjusting the power of the bottom heater of the quartz crucible;

[0171] The time prediction model is trained using a first training sample, wherein the first training sample uses a historical feeding parameter as an input sample and uses a historical adjustment time corresponding to the historical feeding parameter as an output sample;

[0172] The power control unit 903 is used to adjust the power of the bottom heater from a first power value to a second power value at least according to the predicted time parameter to control the temperature of the silicon material in the quartz crucible; the second power value is less than the first power value.

[0173] It can be seen from the above scheme that in an intelligent temperature control device based on data analysis provided in an embodiment of the present application, the relationship between the feeding parameters and the adjustment time in the historical adjustment process of the power of the bottom heater of the quartz crucible is learned by training the model. In this way, the model obtained by training can be used to predict the time of adjusting the power according to the current feeding parameters, so as to adjust the power of the bottom heater according to the predicted time, and there is no need to manually adjust the power according to experience, avoiding inaccurate adjustment caused by insufficient manual experience, thereby improving the accuracy of power adjustment, and then improving the control accuracy of the silicon material temperature in the quartz crucible.

[0174] In one implementation, the data prediction unit 902 is further used to: before the power control unit 903 adjusts the power of the bottom heater from the first power value to the second power value at least according to the prediction time parameter, use the liquid-solid ratio prediction model to process the current feeding parameters to obtain second prediction data; the second prediction data includes: the predicted change rate of the liquid-solid ratio of the silicon material before adjusting the power of the bottom heater; the liquid-solid ratio prediction model is trained using a second training sample, the second training sample uses the historical feeding parameters as input samples, and uses the historical change rate of the liquid-solid ratio of the silicon material before the historical adjustment time as output samples;

[0175] Among them, the power control unit 903 is specifically used to: monitor whether the current change rate of the liquid-to-solid ratio of the silicon material is consistent with the predicted change rate; if the current change rate of the liquid-to-solid ratio of the silicon material is consistent with the predicted change rate and occurs before the starting moment corresponding to the predicted time parameter, at the starting moment corresponding to the predicted time parameter, adjust the power of the bottom heater from the first power value to the second power value; if the current change rate of the liquid-to-solid ratio of the silicon material is consistent with the predicted change rate and occurs within the time range corresponding to the predicted time parameter, at the moment when the current change rate of the liquid-to-solid ratio of the silicon material is consistent with the predicted change rate, adjust the power of the bottom heater from the first power value to the second power value; if within the time range corresponding to the predicted time parameter, the current change rate of the liquid-to-solid ratio of the silicon material is not consistent with the predicted change rate, at the last moment within the time range corresponding to the predicted time parameter, adjust the power of the bottom heater from the first power value to the second power value.

[0176] In one implementation, when the second power value is 0, the data prediction unit 902 is further used to: use a first duration prediction model to process the current feeding parameter and the actual adjustment time of adjusting the power of the bottom heater to obtain a first predicted duration; the first predicted duration represents: starting from when the power of the bottom heater is 0 based on the actual adjustment time, predicting the time required for the liquid-to-solid ratio of the silicon material to reach the first threshold;

[0177] The first duration prediction model is trained using a third training sample, wherein the third training sample uses the historical feeding parameters and the historical adjustment time as input samples, and uses the historical duration required for the silicon material liquid-to-solid ratio to reach a first threshold starting from when the power of the bottom heater is 0 based on the historical adjustment time as an output sample;

[0178] Based on this, the power control unit 903 is also used to: when the first predicted time is reached, determine whether the current liquid-to-solid ratio of the silicon material is less than the first threshold; if the current liquid-to-solid ratio is less than the first threshold, adjust the power of the bottom heater to a third power value; the third power value is greater than 0; monitor whether the shutdown condition is met, and when the shutdown condition is met, turn off the bottom heater.

[0179] The closing condition includes at least one of the following:

[0180] The duration of the bottom heater operating at the third power value reaches a duration threshold;

[0181] The bottom heater is operated at the third power value so that the current liquid-to-solid ratio increases by an increment threshold value;

[0182] The bottom heater is operated at the third power value so that the current liquid-to-solid ratio reaches the first threshold value.

[0183] In one implementation, when the second power value is 0, the data prediction unit 902 is further used to: use a second duration prediction model to process the current feeding parameter and the actual adjustment time of adjusting the power of the bottom heater to obtain a second predicted duration; the second predicted duration represents: starting from making the power of the bottom heater 0 based on the actual adjustment time, predicting the time required for the liquid-to-solid ratio of the silicon material to reach the second threshold;

[0184] The second duration prediction model is trained using a fourth training sample, the fourth training sample uses the historical feeding parameters and the historical adjustment time as input samples, and uses the historical duration required for the silicon material liquid-to-solid ratio to reach a second threshold starting from when the power of the bottom heater is 0 based on the historical adjustment time as an output sample;

[0185] Based on this, the power control unit 903 is also used to: obtain the actual duration of the current liquid-to-solid ratio of the silicon material reaching the second threshold; and adjust the power of the main heater of the quartz crucible according to the actual duration and the second predicted duration.

[0186] Specifically, when the power control unit 903 adjusts the power of the main heater of the quartz crucible according to the actual duration and the second predicted duration, it is specifically used to:

[0187] Determine whether the actual duration is greater than the second predicted duration; if the actual duration is less than the second predicted duration, control the power of the main heater to decrease; if the actual duration is greater than the second predicted duration, control the power of the main heater to increase.

[0188] In one implementation, after the current liquid-to-solid ratio reaches the second threshold, the power control unit 903 is further used to: monitor whether the current liquid-to-solid ratio of the silicon material drops below the second threshold; if the current liquid-to-solid ratio drops below the second threshold, control the power of the main heater to increase until the current liquid-to-solid ratio reaches the second threshold, and return to execute: obtain the actual duration of the current liquid-to-solid ratio of the silicon material reaching the second threshold.

[0189] In one implementation, after the current liquid-to-solid ratio of the silicon material in the quartz crucible reaches 100% and the quartz crucible is positioned, the power control unit 903 is further used to: collect a current image of the silicon material in the quartz crucible; obtain a current temperature value and a temperature change rate of the silicon material in the quartz crucible according to the current image; and adjust the power of the main heater of the quartz crucible according to the current temperature value and the temperature change rate.

[0190] Specifically, when the power control unit 903 adjusts the power of the main heater of the quartz crucible according to the current temperature value and the temperature change rate, it is specifically used to: if the temperature difference between the current temperature value and the target temperature value is greater than or equal to the first difference, and the current temperature value cannot be reduced to match the target temperature value within the target time according to the temperature change rate, control the power of the main heater to be reduced; if the temperature difference between the current temperature value and the target temperature value is greater than or equal to the first difference, and the current temperature value cannot be increased to match the target temperature value within the target time according to the temperature change rate, control the power of the main heater to be increased.

[0191] It should be noted that the specific implementation of each unit in this embodiment can refer to the corresponding content in the previous text and will not be described in detail here.

[0192] The present application also provides an electronic device in an embodiment. Fig.10 As shown, it shows a schematic diagram of the structure of an electronic device suitable for implementing the embodiment of the present application. The electronic device in the embodiment of the present application may include but is not limited to fixed terminals such as mobile phones, laptops, PDAs (personal digital assistants), PADs (tablet computers), desktop computers, etc. Fig.10 The electronic device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0193] like Fig.10As shown, the electronic device may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 1001, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1008 to a random access memory (RAM) 1003. When the electronic device is powered on, various programs and data required for the operation of the electronic device are also stored in the RAM 1003. The processing device 1001, the ROM 1002, and the RAM 1003 are connected to each other via a bus 1004. An input / output (I / O) interface 1005 is also connected to the bus 1004.

[0194] Among them, in this embodiment, the processing device 1001 is used to execute the program in ROM1002 or the program in RAM1003, so that the electronic device can implement any one of the intelligent temperature control methods based on data analysis provided in this embodiment.

[0195] Typically, the following devices may be connected to the I / O interface 1005: an input device 1006 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 1007 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 1008 including, for example, a memory card, a hard disk, etc.; and a communication device 1009. The communication device 1009 may allow the electronic device to communicate with other devices wirelessly or by wire to exchange data. Although Fig.10 An electronic device having various devices is shown, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.

[0196] Also provided in an embodiment of the present application is a computer program product including computer-readable instructions. When the computer-readable instructions are executed on an electronic device, the electronic device implements any one of the intelligent temperature control methods based on data analysis provided in the embodiments of the present application.

[0197] A computer-readable storage medium is also provided in an embodiment of the present application. The storage medium carries one or more computer programs. When one or more computer programs are executed by an electronic device, the electronic device can implement any one of the intelligent temperature control methods based on data analysis provided in the embodiment of the present application.

[0198] The following is an example of the technical solution of this application:

[0199] First, in the traditional single crystal growth control system, it mainly relies on manual real-time observation of the material situation and manual adjustment of the power. There are many abnormal factors such as small observation field, solid material floating on the liquid and the liquid temperature cannot be obtained, and the thickness of the solid material is different. In addition, due to many factors such as different feeding amount of the single crystal furnace each time, different thermal insulation, thermal field thermal insulation attenuation, etc., a set of fixed parameters is not universal.

[0200] Moreover, the current intelligent temperature control method in the industry is still in the exploratory stage, mainly based on real-time observation of CCD images to manually check the proportion and change rate of solid silicon material and liquid silicon material, and combine manual experience to determine whether to adjust the power.

[0201] This application introduces data analysis methods, selects key features in the temperature control process from historical data combined with expert experience, and constructs an effective data set based on the timing of each manual power adjustment, the rate of change of the liquid-to-solid ratio, and the final effect, and uses it as the input of the data analysis model. The classification method is used to classify the effective data set by step length and count the effective intervals, and output the calculated model. The following is a detailed description:

[0202] First of all, the technical problem to be solved by this application is: to use data analysis technology to establish an intelligent temperature control model to provide an automatic temperature control method to solve the problems in the prior art of extended working hours, energy consumption, and increased production costs due to the uncertainty of manual temperature control.

[0203] To solve this technical problem, the solution of this application is as follows:

[0204] First, through the method of data analysis, combined with manual experience, the weight of this re-investment, the weight of each barrel, the time interval between barrels, the rate of change of the liquid-solid ratio before the artificial bottom reduction (i.e., reducing the power of the bottom heater), the bottom reduction (i.e., turning off the bottom heater) and the power adjustment in this temperature control are selected from historical data, and whether the effect of the liquid temperature after this temperature control is within the threshold range (i.e., matching the target temperature value) is used as the standard to further screen out valid data and finally construct a data set (i.e., training sample). Finally, the data set is constructed, and the machine learning gradient boosting decision tree model is used to train the model, and the trained model (i.e., the time prediction model and liquid-solid ratio prediction model mentioned above) is output. The parameters of this temperature control are input into the trained model to predict the time range of the bottom reduction, the liquid-solid ratio change slope (i.e., change rate) before the bottom reduction, the time range of the bottom reduction, and the liquid-solid ratio change slope before the bottom reduction in this temperature control process.

[0205] Secondly, according to the time taken from the historical data to turn off the bottom heating until the liquid-solid ratio reaches threshold 1 (i.e., the first threshold) and threshold 2 (i.e., the second threshold), and without human intervention as the standard, a data set is constructed, and the machine learning gradient boosting decision tree model is used to train the model, and the trained model (i.e., the first duration prediction model and the second duration prediction model mentioned above) is output, and the parameters of this temperature control are input into the trained model to predict the time taken from the bottom heating to the liquid-solid ratio reaching threshold 1 and threshold 2. If the threshold 1 is not reached after a period of time after the bottom heating is turned off, the bottom heater is turned on again, and the liquid-solid ratio is turned off after a certain rate of change or after reaching threshold 1.

[0206] Again, after the bottom addition is turned off and the liquid-to-solid ratio reaches threshold 2, the main addition is linearly adjusted according to the multiples of the seeding power based on the time taken for the bottom addition to reach threshold 2 to prevent overheating.

[0207] At the same time, due to the limited field of view, if the solid silicon material that floats out of the field of view floats back into the field of view, the liquid-to-solid ratio decreases. At this time, the given power of the main heater is increased again. When the liquid-to-solid ratio reaches the threshold value 2 again, the process is repeated.

[0208] Finally, after the silicon material is fully melted and the crucible position is fixed, the real liquid temperature can be obtained through the image grayscale value. At this time, dynamic linear adjustment is performed according to the liquid temperature trend (i.e., the temperature change rate mentioned above) to ensure that the liquid temperature meets the standard after the temperature control is completed.

[0209] The specific steps include:

[0210] 1. Build a model.

[0211] (1) Feature extraction and model building. Fig.11 The flowchart of the model establishment is shown in Figure 1. The re-investment weight, weight of each drum, and time interval between drums at the beginning of each temperature control are extracted from the historical data, and abnormal situations such as manual intervention and unreasonable power adjustment timing are cleared to construct a data set.

[0212] According to different goals, six sub-models are constructed using a machine learning algorithm based on the gradient boosting GBDT (GradientBoosting Decision Tree). After the temperature control starts, each model is called to receive the returned results.

[0213] Model 1: The time frame of bottom-down and top-up;

[0214] Model 2: The slope range of the liquid-to-solid ratio before the bottoming addition is within a certain threshold, such as 10%;

[0215] Model 3: Time range added at the end of the level;

[0216] Model 4: The slope range of the liquid-to-solid ratio before the bottom addition is within a certain threshold, such as 10%;

[0217] Model 5: The time taken to reach the threshold 1 at the bottom of the level.

[0218] Model 6: The time required to reach threshold 2 from the bottom of the level.

[0219] 2. Start temperature control->reduce bottom addition->turn off bottom addition.

[0220] like Fig.12 As shown in the intelligent temperature control logic flow chart, after the temperature control starts, the main heater power is 100kw and the auxiliary heater power is 80kw. According to the pre-conditions of this temperature control, the re-investment weight, the weight of each cylinder, and the time interval between cylinders are input to obtain the inference results of Model 1, Model 2, Model 3, and Model 4.

[0221] When the temperature control meets the reasoning range of both model 1 and model 2, the bottom is reduced and the top is increased.

[0222] When the temperature control meets the reasoning range of both model 3 and model 4, the bottom is turned off.

[0223] 3. Level -> Threshold 1 -> Threshold 2.

[0224] After turning off the bottom heater, continue to monitor the change of liquid-solid ratio. When the duration reaches the inference result of model 5 and the liquid-solid ratio still does not reach threshold 1, the bottom heater is turned on again. After that, when the liquid-solid ratio increases by 10%, reaches threshold 1, or after 10 minutes, the bottom heater is turned off. In this way, the abnormal conditions such as the slow melting speed caused by the shape change of solid blocks can be solved.

[0225] 4. Threshold 2->Full Melt.

[0226] After reaching threshold 2, the liquid-solid ratio changes are continuously monitored. At the same time, according to the ratio of the time taken for the liquid-solid ratio to reach threshold 2 and model 6, the main heater power and duration are linearly set according to the seeding power. The main heater power limit range.

[0227] If the solid silicon material that floated out of the field of view floats back into the field of view due to the limitation of the field of view of the CCD camera in the early stage, causing the liquid-solid ratio to drop to threshold 3 (threshold 3 is less than threshold 2), the main heater power is reset to a multiple of the seeding power. When the liquid-solid ratio reaches threshold 2 again, repeat this process.

[0228] 5. Fully melted -> Temperature control completed.

[0229] After the silicon material is fully melted and the crucible position is set, the actual liquid temperature is obtained, and the deviation from the target liquid temperature of 1450℃±2℃ and the change trend of the liquid temperature in the first 10 cycles (i.e., the temperature change rate) are dynamically adjusted linearly to ensure that the liquid temperature meets the standard after the temperature control is completed.

[0230] It can be seen that this application uses data analysis methods and combines manual experience to select key features from the beginning to the end of temperature control from historical data, and further screens out effective features based on the current re-investment weight, power adjustment timing, liquid-to-solid ratio change rate, and whether the liquid temperature after the temperature control is in a reasonable range as standards. These effective features are classified and statistically calculated to obtain an effective range, and the timing of reducing the bottom power and shutting down the bottom power next time can be preliminarily deduced. At the same time, the slope of the change in the liquid-to-solid ratio is monitored in real time to infer whether the bottom heater needs to be reopened or the main heater needs to be lowered, how long it lasts, and how much power is given. This application establishes an intelligent temperature control model through data analysis methods to solve the uncertainty of manual temperature control in the prior art.

[0231] In summary, the technical solution of the present application can improve the automated decision-making ability in the process of pulling single crystal silicon and reduce the possibility of manual intervention and misjudgment. Moreover, the technical solution of the present application can solve the problem of manual power adjustment in the process of pulling single crystal silicon, so that the solid silicon material can be converted into liquid silicon material in the shortest time and with the least power, and the liquid temperature needs to ensure consistency. Therefore, the present application can not only intelligently control the temperature and improve the degree of automation, but also effectively reduce labor costs, energy consumption, extend working hours, and improve production efficiency.

[0232] It should also be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed over multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the drawings of the device embodiments provided by the present application, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines.

[0233] Through the description of the above implementation mode, the technicians in the field can clearly understand that the present application can be implemented by means of software plus necessary general hardware, and of course, it can also be implemented by special hardware including special integrated circuits, special CPUs, special memories, special components, etc. In general, all functions completed by computer programs can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be various, such as analog circuits, digital circuits or special circuits. However, for the present application, software program implementation is a better implementation mode in more cases. Based on such an understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a readable storage medium, such as a computer floppy disk, a U disk, a mobile hard disk, a ROM, a RAM, a disk or an optical disk, etc., including a number of instructions to enable a computer device (which can be a personal computer, a training device, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0234] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments may be implemented in the form of a computer program product.

[0235] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from a website site, a computer, a training device, or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, training device, or data center. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a training device, a data center, etc. that includes one or more available media integrations. The available medium may be a magnetic medium, (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)), etc.

Claims

1. An intelligent temperature control method based on data analysis, It is characterized in that The method comprises: Obtaining current feeding parameters for re-feeding a quartz crucible used for producing single crystal silicon; The current feeding parameters are processed by using a time prediction model to obtain first prediction data; the first prediction data includes: a prediction time parameter for adjusting the power of the bottom heater of the quartz crucible; The time prediction model is trained using a first training sample, wherein the first training sample uses a historical feeding parameter as an input sample and uses a historical adjustment time corresponding to the historical feeding parameter as an output sample; At least according to the predicted time parameter, adjusting the power of the bottom heater from a first power value to a second power value to control the temperature of the silicon material in the quartz crucible, wherein the second power value is less than the first power value; Before adjusting the power of the bottom heater from the first power value to the second power value according to at least the predicted time parameter, the method further comprises: The current feeding parameters are processed by using a liquid-to-solid ratio prediction model to obtain second prediction data; the second prediction data includes: a predicted change rate of the liquid-to-solid ratio of the silicon material before adjusting the power of the bottom heater; the liquid-to-solid ratio prediction model is trained by using a second training sample, the second training sample uses historical feeding parameters as input samples, and uses a historical change rate of the liquid-to-solid ratio of the silicon material before the historical adjustment time as an output sample; Wherein, adjusting the power of the bottom heater from a first power value to a second power value at least according to the predicted time parameter comprises: Monitoring whether the current change rate of the liquid-to-solid ratio of the silicon material is consistent with the predicted change rate; If the current change rate of the liquid-to-solid ratio of the silicon material is consistent with the predicted change rate and occurs before the start time corresponding to the predicted time parameter, at the start time corresponding to the predicted time parameter, the power of the bottom heater is adjusted to switch from the first power value to the second power value; If the current change rate of the liquid-to-solid ratio of the silicon material is consistent with the predicted change rate within the time range corresponding to the predicted time parameter, at the moment when the current change rate of the liquid-to-solid ratio of the silicon material is consistent with the predicted change rate, adjusting the power of the bottom heater to switch from the first power value to the second power value; If within the time range corresponding to the predicted time parameter, the current change rate of the liquid-to-solid ratio of the silicon material is not consistent with the predicted change rate, at the last moment within the time range corresponding to the predicted time parameter, the power of the bottom heater is adjusted to switch from the first power value to the second power value.

2. The method according to claim 1, It is characterized in that When the second power value is 0, the method further includes: The current feeding parameters and the actual adjustment time of adjusting the power of the bottom heater are processed by using a first duration prediction model to obtain a first prediction duration; the first prediction duration represents: starting from when the power of the bottom heater is made 0 based on the actual adjustment time, the time required for the liquid-to-solid ratio of the silicon material to reach a first threshold is predicted; The first duration prediction model is trained using a third training sample, wherein the third training sample uses the historical feeding parameters and the historical adjustment time as input samples, and uses the historical duration required for the silicon material liquid-to-solid ratio to reach a first threshold starting from when the power of the bottom heater is 0 based on the historical adjustment time as an output sample; When the first prediction time is reached, determining whether the current liquid-to-solid ratio of the silicon material is less than the first threshold; If the current liquid-to-solid ratio is less than the first threshold, adjusting the power of the bottom heater to a third power value; the third power value is greater than 0; It is monitored whether a shut-down condition is satisfied, and when the shut-down condition is satisfied, the bottom heater is shut down.

3. The method according to claim 2, It is characterized in that The closing condition includes at least one of the following: The duration of the bottom heater operating at the third power value reaches a duration threshold; The bottom heater is operated at the third power value so that the current liquid-to-solid ratio increases by an increment threshold value; The bottom heater is operated at the third power value so that the current liquid-to-solid ratio reaches the first threshold value.

4. The method according to claim 1, It is characterized in that When the second power value is 0, the method further includes: The second duration prediction model is used to process the current feeding parameters and the actual adjustment time of adjusting the power of the bottom heater to obtain a second predicted duration; the second predicted duration represents: starting from when the power of the bottom heater is made 0 based on the actual adjustment time, the time required for the liquid-to-solid ratio of the silicon material to reach the second threshold is predicted; The second duration prediction model is trained using a fourth training sample, the fourth training sample uses the historical feeding parameters and the historical adjustment time as input samples, and uses the historical duration required for the silicon material liquid-to-solid ratio to reach a second threshold starting from when the power of the bottom heater is 0 based on the historical adjustment time as an output sample; Obtaining the actual duration of time during which the current liquid-to-solid ratio of the silicon material reaches the second threshold; The power of the main heater of the quartz crucible is adjusted according to the actual duration and the second predicted duration.

5. The method according to claim 4, It is characterized in that According to the actual duration and the second predicted duration, adjusting the power of the main heater of the quartz crucible includes: Determining whether the actual duration is greater than the second predicted duration; If the actual duration is less than the second predicted duration, controlling the power of the main heater to decrease; If the actual duration is greater than the second predicted duration, the power of the main heater is controlled to increase.

6. The method according to claim 4, It is characterized in that After the current liquid-to-solid ratio reaches the second threshold, the method further includes: Monitoring whether the current liquid-to-solid ratio of the silicon material drops below the second threshold; If the current liquid-to-solid ratio drops below the second threshold, the power of the main heater is controlled to increase until the current liquid-to-solid ratio reaches the second threshold, and the process returns to execute: obtaining the actual duration of the current liquid-to-solid ratio of the silicon material reaching the second threshold.

7. The method according to claim 1, It is characterized in that After the current liquid-to-solid ratio of the silicon material in the quartz crucible reaches 100% and the quartz crucible is positioned, the method further includes: Acquiring a current image of the silicon material in the quartz crucible; According to the current image, obtaining the current temperature value and temperature change rate of the silicon material in the quartz crucible; The power of the main heater of the quartz crucible is adjusted according to the current temperature value and the temperature change rate.

8. The method according to claim 7, It is characterized in that According to the current temperature value and the temperature change rate, adjusting the power of the main heater of the quartz crucible includes: If the temperature difference between the current temperature value and the target temperature value is greater than or equal to a first difference, and the current temperature value cannot be reduced to match the target temperature value within a target time period according to the temperature change rate, controlling the power of the main heater to be reduced; If the temperature difference between the current temperature value and the target temperature value is greater than or equal to the first difference, and the current temperature value cannot be increased to match the target temperature value within the target time period according to the temperature change rate, the power of the main heater is controlled to increase.

9. An intelligent temperature control device based on data analysis, It is characterized in that The device comprises: A parameter acquisition unit, used to obtain current feeding parameters for re-feeding a quartz crucible used for producing single crystal silicon; A data prediction unit is used to process the current feeding parameters using a time prediction model to obtain first prediction data; the first prediction data includes: a prediction time parameter for adjusting the power of the bottom heater of the quartz crucible; The time prediction model is trained using a first training sample, wherein the first training sample uses a historical feeding parameter as an input sample and uses a historical adjustment time corresponding to the historical feeding parameter as an output sample; A power control unit, configured to adjust the power of the bottom heater from a first power value to a second power value at least according to the predicted time parameter, so as to control the temperature of the silicon material in the quartz crucible; the second power value is less than the first power value; The data prediction unit is further used for: before the power control unit adjusts the power of the bottom heater from the first power value to the second power value at least according to the prediction time parameter, using the liquid-solid ratio prediction model to process the current feeding parameters to obtain second prediction data; the second prediction data includes: the predicted change rate of the liquid-solid ratio of the silicon material before adjusting the power of the bottom heater; the liquid-solid ratio prediction model is trained using a second training sample, the second training sample uses the historical feeding parameters as input samples, and uses the historical change rate of the liquid-solid ratio of the silicon material before the historical adjustment time as output samples; The power control unit is specifically used to: monitor whether the current change rate of the liquid-to-solid ratio of the silicon material is consistent with the predicted change rate; if the current change rate of the liquid-to-solid ratio of the silicon material is consistent with the predicted change rate and occurs before the starting moment corresponding to the predicted time parameter, at the starting moment corresponding to the predicted time parameter, adjust the power of the bottom heater from the first power value to the second power value; if the current change rate of the liquid-to-solid ratio of the silicon material is consistent with the predicted change rate and occurs within the time range corresponding to the predicted time parameter, at the moment when the current change rate of the liquid-to-solid ratio of the silicon material is consistent with the predicted change rate, adjust the power of the bottom heater from the first power value to the second power value; if within the time range corresponding to the predicted time parameter, the current change rate of the liquid-to-solid ratio of the silicon material is not consistent with the predicted change rate, at the last moment within the time range corresponding to the predicted time parameter, adjust the power of the bottom heater from the first power value to the second power value.

Images