Single crystal drawing temperature control method and device and storage medium

Abstract

The invention relates to the technical field of single crystal silicon production, and discloses a single crystal drawing control method and device and a storage medium, and the method comprises the following steps: solving a temperature set value sequence with the lowest loss rate according to single crystal drawing historical data by using a crowd-sourcing algorithm; training a single crystal drawing temperature control strategy model by using single crystal drawing historical data; solving a temperature control strategy by using the single crystal drawing temperature control strategy model, so that the temperature in the single crystal furnace changes according to a temperature set value sequence; and performing closed-loop control on the heating system and the cooling system according to the temperature control strategy. According to the method, the yield of monocrystalline silicon can be improved, the control quality of the temperature in the single crystal furnace is improved, and the uniformity of the quality of the produced monocrystalline silicon is improved. According to the method, the quality of the produced monocrystalline silicon can be prevented from being limited by experience of operators, and the quality of the produced monocrystalline silicon can be stably controlled.

Description

Technical Field

[0001] The present invention relates to the technical field of single crystal silicon production, and in particular to a single crystal pulling temperature control method, device and storage medium. Background Art

[0002] With the development of photovoltaic technology and 5G technology, the market demand for monocrystalline silicon continues to increase. The growth process of monocrystalline silicon has extremely high requirements for temperature control. In the existing single crystal furnace growth process, temperature regulation mainly relies on manual intervention, resulting in production quality relying on the experience of the operator, and the quality of the produced monocrystalline silicon cannot be stably controlled. Therefore, it is necessary to design a new single crystal pulling temperature control scheme that can stably control the production quality of monocrystalline silicon. Summary of the invention

[0003] In order to overcome the deficiencies of the prior art, the object of the present invention is to provide a single crystal pulling control method that can improve the control quality of the temperature in the single crystal furnace, reduce temperature fluctuations, and improve the yield and quality uniformity of the single crystal.

[0004] To solve the above problems, the technical solution adopted by the present invention is as follows: A single crystal pulling temperature control method comprises the following steps:

[0005] Using the crowd intelligence algorithm, according to the historical data of single crystal pulling, the temperature setting value sequence with the lowest loss rate is solved;

[0006] Use the historical data of single crystal pulling to train the single crystal pulling temperature control strategy model;

[0007] The temperature control strategy model of single crystal pulling is used to solve the temperature control strategy, so that the temperature in the single crystal furnace changes according to the temperature setting value sequence;

[0008] The heating and cooling systems are closed-loop controlled according to the temperature control strategy.

[0009] Compared with the prior art, the beneficial effect of the present invention is that: this method obtains experience from historical single crystal pulling data to obtain a sequence of temperature setting values ​​that can control the loss rate of single crystal pulling to the lowest, thereby controlling the temperature in the single crystal furnace to change with a more optimal temperature curve, thereby improving the yield of single crystals. At the same time, by using the single crystal pulling temperature control strategy model trained by historical data to determine the temperature control strategy according to actual conditions, the temperature fluctuation in the single crystal furnace can be reduced, the quality of temperature control in the single crystal furnace can be improved, and the uniformity of the quality of the produced single crystals can be improved. This method uses the temperature control data of historical single crystal pulling as experience to guide the control of single crystal pulling temperature, which can avoid the quality of the produced single crystal silicon being restricted by the experience of the operator.

[0010] In the above-mentioned single crystal pulling temperature control method, in the step of using a swarm intelligence algorithm to solve a temperature setting value sequence with the lowest loss rate based on the single crystal pulling historical data, a genetic algorithm is used to solve the temperature setting value sequence with the lowest loss rate.

[0011] In the above-mentioned single crystal pulling temperature control method, the step of using the crowd intelligence algorithm to solve the temperature setting value sequence with the lowest loss rate according to the single crystal pulling historical data specifically includes:

[0012] Obtain a certain number of single crystal pulling historical temperature sequences and corresponding defect rates, and use the single crystal pulling historical temperature sequences and corresponding defect rates to train a defect rate prediction model;

[0013] Encode each single crystal pulling history temperature sequence as a chromosome;

[0014] A certain number of chromosomes are randomly selected to form an initial population. The fitness of each chromosome is calculated according to the defect rate of each chromosome in the initial population, and the chromosomes with higher fitness are selected as parents.

[0015] Cross the parental chromosomes in pairs to generate a new chromosome population, and randomly mutate the chromosomes in the new chromosome population;

[0016] The defect rate prediction model is used to predict the defect rate of each chromosome in the new chromosome population;

[0017] The process of calculating chromosome fitness, selecting parents, crossover, and mutation is repeated until the preset number of iterations is reached;

[0018] The chromosome with the highest fitness in the chromosome population after the last iteration is selected as the temperature setting value sequence with the lowest loss rate.

[0019] In the above-mentioned single crystal pulling temperature control method, in the step of using the single crystal pulling historical data to train the single crystal pulling temperature control strategy model, the DQN algorithm is used to train the single crystal pulling temperature control strategy model.

[0020] In the above-mentioned single crystal pulling temperature control method, the step of using the single crystal pulling historical data to train the single crystal pulling temperature control strategy model specifically includes:

[0021] Initialize the weights of the Q network, the target Q network, and the experience pool;

[0022] Define state space, action space, and reward function;

[0023] Acquire single crystal pulling history data, and store the acquired single crystal pulling history data as experience in an experience pool according to a preset format;

[0024] Sample a batch of experience from the experience pool, and update the weight of the Q network through the TD algorithm based on the sampled experience;

[0025] After a certain number of iterations, the weight of the Q network is updated to the target Q network.

[0026] A storage medium stores a computer program, wherein the computer program is configured to implement the above-mentioned single crystal pulling temperature control method when called and executed by a processor.

[0027] A single crystal pulling temperature control device comprises a processor and a memory, wherein the processor is electrically connected to the memory, and the processor can implement the above-mentioned single crystal pulling temperature control method by calling and executing a computer program in the memory.

[0028] A single crystal pulling temperature control system comprises: the above-mentioned single crystal pulling temperature control device; a heating system, controlled by the single crystal pulling temperature control device, and used for heating the environment in the single crystal furnace; a cooling system, controlled by the single crystal pulling temperature control device, and used for quickly adjusting and maintaining the temperature in the single crystal furnace; a temperature sensing network, communicatively connected to the single crystal pulling temperature control device, comprising a plurality of temperature sensors, wherein the plurality of temperature sensors are distributed at the top, middle and bottom of the single crystal furnace.

[0029] The above-mentioned single crystal pulling temperature control system, the heating system includes a heating device and a heating drive mechanism, the heating device is arranged on the heating drive mechanism, and the heating drive mechanism is used to drive the heating device to move in the single crystal furnace.

[0030] The above-mentioned single crystal pulling temperature control system also includes an integrated controller and a communication module. The communication module, the heating system, the cooling system and the temperature sensing network are all electrically connected to the integrated controller. The integrated controller is communicatively connected to the single crystal pulling temperature control device through the communication module.

[0031] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a flow chart of a single crystal pulling temperature control method according to an embodiment of the present invention;

[0033] Figure 2 A flowchart of a process for solving an optimal temperature setting value sequence according to an embodiment of the present invention;

[0034] Figure 3 It is a flow chart of the training process of the single crystal pulling temperature control strategy model according to an embodiment of the present invention;

[0035] Figure 4It is a schematic structural diagram of a single crystal pulling temperature control system according to an embodiment of the present invention.

[0036] Description of Figure Numbers:

[0037] 100 heating system, 110 heating device, 120 heating drive mechanism, 200 cooling system, 300 temperature sensor. DETAILED DESCRIPTION

[0038] The embodiments of the present invention are described in detail below. Figure 1 , an embodiment of the present invention provides a single crystal pulling temperature control method, comprising the following steps:

[0039] Using the crowd intelligence algorithm, according to the historical data of single crystal pulling, the temperature setting value sequence with the lowest loss rate is solved;

[0040] Use the historical data of single crystal pulling to train the single crystal pulling temperature control strategy model;

[0041] The temperature control strategy model of single crystal pulling is used to solve the temperature control strategy, so that the temperature in the single crystal furnace changes according to the temperature setting value sequence;

[0042] The heating system 100 and the cooling system 200 are closed-loop controlled according to the temperature control strategy.

[0043] The method obtains experience from historical single crystal pulling data, uses historical data as a guide to solve the temperature setting value sequence with the lowest loss rate through a group intelligence algorithm, and sets the temperature of the single crystal furnace at each time node according to the solved temperature setting value sequence, so that the temperature in the single crystal furnace can be controlled to change according to the temperature curve with the lowest loss rate, so as to control the loss rate of single crystal silicon within a lower range and improve the yield rate of single crystal. At the same time, the method also determines the temperature control strategy according to actual conditions by using the single crystal pulling temperature control strategy model trained by historical data, which can reduce the temperature fluctuation in the single crystal furnace, improve the control quality of the temperature in the single crystal furnace, and improve the uniformity of the quality of the produced single crystal. The method uses the temperature control data of historical single crystal pulling as experience to guide the control of the temperature in the single crystal pulling process, which can avoid the production quality of the single crystal being overly dependent on the experience of the operator, can stably control the quality of the produced single crystal silicon, can automatically control the temperature in the single crystal furnace, and reduce the training cost for the operator.

[0044] It is understandable that the swarm intelligence algorithm can adopt a particle swarm algorithm, a genetic algorithm or an ant colony algorithm. Figure 2 In this embodiment, a genetic algorithm is used to solve the temperature setting value sequence with the lowest loss rate. Taking the target temperature value set once every minute and solving the temperature setting value sequence of ten minutes each time as an example, the specific steps include the following:

[0045] First, historical single crystal pulling production data is collected to form a historical data set, wherein the single crystal pulling production data includes parameters such as temperature setting value, actual temperature and defect rate per minute, and then the sequence of temperature setting value of each single crystal pulling process in the historical data set is divided into multiple historical temperature sequences, each historical temperature sequence includes temperature setting values ​​for ten consecutive minutes;

[0046] Then, the historical temperature sequence is encoded according to a preset encoding format to form multiple chromosomes, for example, chromosome = [T1, T2, T3, ..., T10], where T1 to T10 are the temperature setting values ​​per minute in each chromosome;

[0047] A certain number of chromosomes are randomly selected to form an initial population. The fitness of each chromosome in the initial population is calculated according to the defect rate of each chromosome, and the chromosome with higher fitness is selected as the parent. It can be understood that the value of fitness should be negatively correlated with the defect rate. In this embodiment, fitness = 1 / (defect_rate + 1e -6 ), where defect_rate is the defect rate of the historical temperature sequence corresponding to each chromosome. It can be understood that parents can be selected from the initial population by selection operators such as random selection, fitness ratio selection, elitism, and tournament selection. In this embodiment, 100 chromosomes are randomly selected as the initial population, and chromosomes with higher fitness are selected as parents by roulette selection method;

[0048] Then, the selected parents are crossed in pairs to form a new chromosome group, and the chromosomes in the new chromosome group are randomly mutated. It can be understood that the crossover operator and the mutation operator can be selected according to actual needs. In this embodiment, a single-point crossover method is used to generate new offspring, and a gene of the chromosome is randomly changed with a probability of 10%.

[0049] Since there is no corresponding defect rate record for the chromosomes in the newly generated population, it is necessary to use a defect rate prediction model trained by a historical data set to predict the defect rate of the chromosomes in the newly generated chromosome population, so as to evaluate the fitness of the new chromosomes according to the predicted defect rate. The defect rate prediction model can use a hierarchical analysis method, a TOPSI S method, or a neural network model. In this embodiment, the historical temperature sequence, the pulling speed, and the rotation speed in the historical single crystal pulling production data are used as inputs, and the defect rate is used as output to train the neural network model to obtain a defect rate prediction model, and the temperature setting value sequence corresponding to the chromosome and the actual values ​​of other production parameters are input into the defect rate prediction model to predict the defect rate of each new chromosome.

[0050] The selection, crossover and mutation process is repeated for the new chromosome population until the number of iterations reaches the preset maximum number of iterations. The chromosome with the highest fitness in the new chromosome population after the last iteration is regarded as the temperature setting value sequence with the lowest defect rate.

[0051] It is understandable that the temperature control strategy model can be obtained by using methods such as long short-term memory network (LSTM), deep neural network (DNN) or reinforcement learning (RL). Figure 3 In this embodiment, a DQN network is used to construct a temperature control strategy model, which specifically includes the following steps:

[0052] Initialize the weights of the Q network and the target Q network, and initialize the experience pool. Initializing the experience pool means setting aside a portion of space to store experience;

[0053] Define the state space, action space and reward function. It can be understood that the state space includes the current temperature, historical temperature, set temperature, pulling speed and rotation speed, etc., and the action space can be the power of adjusting the heating system 100 and the cooling system 200, etc. In this embodiment, since the temperature sensing network for detecting the temperature in the single crystal furnace includes a plurality of temperature sensors 300, and the plurality of temperature sensors 300 are distributed at the top, middle and bottom of the single crystal furnace, the current temperature and historical temperature should include the current temperature and historical temperature of the top, middle and bottom. The heating system 100 includes a heating device 110 and a heating drive mechanism 120, and the heating drive mechanism 120 is used to drive the heating device 110 to move in the single crystal furnace to drive the heating device 110 to heat different areas in the single crystal furnace. Therefore, the action space may include increasing the power of the heating device 110 by a step value, reducing the power of the heating device 110 by a step value, keeping the power of the heating device 110 unchanged, raising the heating device 110 by a step, lowering the heating device 110 by a step, keeping the position of the heating device 110 inconvenient, increasing the power of the cooling device by a step value, reducing the power of the cooling device by a step value, and keeping the power of the cooling device unchanged. The reward function should encourage actions that can make the current temperature close to the set temperature. In this embodiment, when the absolute value of the difference between the current temperature after the action and the set temperature is within the preset range, the action is given a positive reward value of 1; when the absolute value of the difference between the current temperature after the action and the set temperature is outside the preset range, the action is given a negative reward value of -14;

[0054] Obtain the single crystal pulling history data, and store the obtained single crystal pulling history data as experience in the experience pool according to the preset format. Each experience should include the current state, the control action adopted, the state after the action, and the reward value of the action;

[0055] Then a batch of experience is sampled from the experience pool, and the weights of the Q network are updated through the TD algorithm based on the sampled experience;

[0056] At every certain number of iterations, the weight of the Q network is updated to the target Q network. When solving the temperature control strategy, the current state and temperature setting value of the single crystal furnace are input into the latest target Q network to solve the optimal temperature control action, and the heating system 100 and the cooling system 200 are closed-loop controlled according to the solved temperature control action.

[0057] It can be understood that after each single crystal pulling production, the single crystal pulling production data can be recorded in the storage medium, so that the defect rate prediction model and the temperature control strategy model can continue to learn the continuously updated single crystal pulling historical data, so as to improve the accuracy of the defect rate prediction model in predicting the defect rate, and improve the control effect of the temperature control strategy output by the single crystal pulling temperature control strategy model on the temperature of the single crystal furnace.

[0058] Based on the same inventive concept, an embodiment of the present invention further provides a computer-readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, the above-mentioned single crystal pulling control method can be implemented.

[0059] In some possible embodiments, various aspects of the single crystal pulling control method provided by the present invention can also be implemented in the form of a program product, which includes a program code. When the program product is run on an apparatus, the program code is used to enable the control device to execute the steps of the single crystal pulling control method according to various exemplary embodiments of the present application described above in this specification.

[0060] Based on the same inventive concept, an embodiment of the present invention also provides a control device for implementing the above-mentioned single crystal pulling control method, including a processor and a memory, the memory is electrically connected to the processor, and the processor is used to execute the computer program stored in the memory to implement the above-mentioned single crystal pulling control method.

[0061] In one possible design, the processor may include one or more processing units, and the processor and memory may be implemented on the same chip or on separate chips. The processor may be a general-purpose processor, such as a central processing unit (CPU), a digital signal processor, an application-specific integrated circuit, a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component, which may implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present application. A general-purpose processor may be a microprocessor or any conventional processor, etc. The steps of the single crystal pulling control method disclosed in the embodiments of the present application may be directly embodied as being executed by a hardware processor, or may be executed by a combination of hardware and software modules in the processor.

[0062] As a non-volatile computer-readable storage medium, the memory can be used to store non-volatile software programs, non-volatile computer executable programs and modules. The memory can include at least one type of storage medium, such as flash memory, hard disk, multimedia card, card-type memory, random access memory (Random Access Memory, RAM), static random access memory (Static Random Access Memory, SRAM), programmable read-only memory (Programmab le Read On ly Memory, PROM), read-only memory (Read On ly Memory, ROM), electrically erasable programmable read-only memory (Electrically Erasable Programmab le Read-Only Memory, EEPROM), magnetic storage, disk, optical disk, etc. The memory is any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory in the embodiment of the present application can also be a circuit or any other device that can realize a storage function, for storing program instructions and / or data.

[0063] By designing and programming the processor, the code corresponding to the single crystal pulling control method described in the above embodiment can be fixed into the chip, so that the chip can execute the steps of the single crystal pulling control method of the embodiment shown in the present invention when running. How to design and program the processor is a technology well known to those skilled in the art and will not be described in detail here.

[0064] Reference Figure 4 Based on the same inventive concept, an embodiment of the present invention also provides a single crystal pulling temperature control system, including the above-mentioned single crystal pulling temperature control device, a heating system 100 and a cooling system 200 controlled by the single crystal pulling temperature control device, and a temperature sensing network communicatively connected to the single crystal pulling temperature control device. The heating system 100 is used to heat the environment in the single crystal furnace under the control of the single crystal pulling temperature control device, and the cooling system 200 is used to cool the environment in the single crystal furnace under the control of the single crystal pulling temperature control device, so as to achieve rapid adjustment or maintenance of the temperature of the environment in the single crystal furnace. The temperature sensing network includes a plurality of temperature sensors 300, and the plurality of temperature sensors 300 are distributed at the top, middle and bottom of the single crystal furnace, and are used to feed back the temperature of the top, middle and bottom of the single crystal furnace to the single crystal pulling temperature control device.

[0065] Reference Figure 4In some embodiments, the heating system 100 includes a heating device 110 and a heating drive mechanism 120. The heating device 110 is arranged on the heating drive mechanism 120. The heating drive mechanism 120 is used to drive the heating device 110 to move in the single crystal furnace, so as to drive the heating device 110 to heat different areas in the single crystal furnace to adjust the temperature in the single crystal furnace and the uniformity of the temperature field in the single crystal furnace. The heating drive mechanism 120 can be a high-temperature resistant linear drive mechanism. In this embodiment, the heating drive mechanism 120 is vertically arranged in the single crystal furnace, and can drive the heating device 110 to move in the vertical direction, so as to adjust the temperature field in the single crystal furnace to be uniform in the vertical direction. It can be understood that the heating device 110 is made of an electric heating material with a fast response speed, such as metal ceramics. The cooling system 200 can cool the temperature in the single crystal furnace by water cooling or air cooling. In this embodiment, the cooling system 200 is a water-cooled screen surrounding the crucible in the single crystal furnace. The temperature sensor 300 can be a thermocouple or an infrared thermometer.

[0066] In this embodiment, the single crystal pulling control system also includes an integrated controller and a communication module. The communication module, the heating system 100, the cooling system 200 and the temperature sensor network are all electrically connected to the integrated controller. The integrated controller is connected to the single crystal pulling temperature control device through the communication module. The communication module can be an Ethernet module or a Wifi module. The single crystal pulling control device transmits the control instructions of the cooling system 200 and the heating system 100 to the integrated controller through the communication module. The integrated controller controls the cooling system 200 and the heating system 100 to perform corresponding control actions according to the control instructions, so that the temperature in the single crystal furnace is controlled according to the temperature setting value sequence with the lowest loss rate solved by the single crystal pulling control device. At the same time, the integrated controller transmits the temperature in the single crystal furnace and other production parameters collected by the temperature sensor network to the single crystal pulling temperature control device through the communication module, so that the single crystal furnace pulling temperature control device solves the optimal temperature control strategy through the single crystal pulling temperature control strategy model.

[0067] The present application is described with reference to the flowcharts and / or block diagrams of the methods, apparatus (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0068] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0069] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0070] It should be noted that in the description of the present invention, if there are any orientation descriptions, such as up, down, front, back, left, right, etc., the orientations or positional relationships indicated are all based on the orientations or positional relationships shown in the drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed or operated in a specific orientation, and cannot be understood as a limitation on the present invention.

[0071] In the description of the present invention, "several" means one or more, "more" means two or more, "greater than", "less than", "exceed" etc. are understood to exclude the number itself, and "above", "below", "within" etc. are understood to include the number itself. If there is a description of "first" or "second", it is only used to distinguish the technical features, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features or implicitly indicating the order of the indicated technical features.

[0072] In the description of the present invention, unless otherwise clearly defined, terms such as setting, installing, connecting, etc. should be understood in a broad sense, and technicians in the relevant technical field can reasonably determine the specific meanings of the above terms in the present invention based on the specific content of the technical solution.

[0073] The above-mentioned embodiments are only preferred embodiments of the present invention and cannot be used to limit the scope of protection of the present invention. Any non-substantial changes and substitutions made by technicians in this field on the basis of the present invention shall fall within the scope of protection required by the present invention.

Claims

1. A single crystal pulling temperature control method, characterized in that: The steps include: Using the crowd intelligence algorithm, according to the historical data of single crystal pulling, the temperature setting value sequence with the lowest loss rate is solved; Use the historical data of single crystal pulling to train the single crystal pulling temperature control strategy model; The temperature control strategy model of single crystal pulling is used to solve the temperature control strategy, so that the temperature in the single crystal furnace changes according to the temperature setting value sequence; The heating system (100) and the cooling system (200) are closed-loop controlled according to the temperature control strategy.

2. The single crystal pulling temperature control method according to claim 1, characterized in that: In the step of using a swarm intelligence algorithm to solve a temperature setting value sequence with the lowest loss rate based on the single crystal pulling historical data, a genetic algorithm is used to solve the temperature setting value sequence with the lowest loss rate.

3. The single crystal pulling temperature control method according to claim 2, characterized in that: The step of using the swarm intelligence algorithm to solve the temperature setting value sequence with the lowest loss rate according to the single crystal pulling historical data specifically includes: Obtain a certain number of single crystal pulling historical temperature sequences and corresponding defect rates, and use the single crystal pulling historical temperature sequences and corresponding defect rates to train a defect rate prediction model; Encode each single crystal pulling history temperature sequence as a chromosome; A certain number of chromosomes are randomly selected to form an initial population. The fitness of each chromosome is calculated according to the defect rate of each chromosome in the initial population, and the chromosomes with higher fitness are selected as parents. Cross the parental chromosomes in pairs to generate a new chromosome population, and randomly mutate the chromosomes in the new chromosome population; The defect rate prediction model is used to predict the defect rate of each chromosome in the new chromosome population; The process of calculating chromosome fitness, selecting parents, crossover, and mutation is repeated until the preset number of iterations is reached; The chromosome with the highest fitness in the chromosome population after the last iteration is selected as the temperature setting value sequence with the lowest loss rate.

4. The single crystal pulling temperature control method according to claim 1, characterized in that: In the step of using the single crystal pulling historical data to train the single crystal pulling temperature control strategy model, the DQN algorithm is used to train the single crystal pulling temperature control strategy model.

5. The single crystal pulling temperature control method according to claim 4, characterized in that: The step of using the single crystal pulling historical data to train the single crystal pulling temperature control strategy model specifically includes: Initialize the weights of the Q network, the target Q network, and the experience pool; Define state space, action space, and reward function; Acquire single crystal pulling history data, and store the acquired single crystal pulling history data as experience in an experience pool according to a preset format; Sample a batch of experience from the experience pool, and update the weight of the Q network through the TD algorithm based on the sampled experience; After a certain number of iterations, the weight of the Q network is updated to the target Q network.

6. A storage medium storing a computer program, characterized in that: The computer program is configured to implement the single crystal pulling temperature control method according to any one of claims 1 to 5 when called and executed by a processor.

7. A single crystal pulling temperature control device, characterized in that: It comprises a processor and a memory, the processor is electrically connected to the memory, and the processor can implement the single crystal pulling temperature control method according to any one of claims 1 to 5 by calling and executing a computer program in the memory.

8. A single crystal pulling temperature control system, characterized in that: include: The single crystal pulling temperature control device according to claim 7; A heating system (100), controlled by the single crystal pulling temperature control device, for heating the environment in the single crystal furnace; A cooling system (200), controlled by the single crystal pulling temperature control device, for quickly adjusting and maintaining the temperature in the single crystal furnace; The temperature sensing network is communicatively connected to the single crystal pulling temperature control device and comprises a plurality of temperature sensors (300), wherein the plurality of temperature sensors (300) are distributed at the top, middle and bottom of the single crystal furnace.

9. The single crystal pulling temperature control system according to claim 8, characterized in that: The heating system (100) comprises a heating device (110) and a heating drive mechanism (120); the heating device (110) is arranged on the heating drive mechanism (120); and the heating drive mechanism (120) is used to drive the heating device (110) to move in a single crystal furnace.

10. The single crystal pulling temperature control system according to claim 8, characterized in that: It also includes an integrated controller and a communication module, wherein the communication module, the heating system (100), the cooling system (200) and the temperature sensing network are all electrically connected to the integrated controller, and the integrated controller is communicatively connected to the single crystal pulling temperature control device via the communication module.

Images