Continuous graphitization furnace temperature control method and system

Abstract

The invention discloses a temperature control method and system for a continuous graphitization furnace, the graphitization furnace comprises a furnace body and a pair of graphite electrodes inserted in the furnace body, and the temperature control method comprises the following steps: real-time temperature of a material in the furnace body and instantaneous power of a circuit where the graphite electrodes are located are detected in real time; calculating a temperature change rate based on the real-time temperature, and calculating a power change rate based on the instantaneous power; and the intelligent model configured based on the neural network architecture processes the real-time temperature, the temperature change rate and the power change rate to obtain a first adjustment value used for adjusting the output power of the circuit where the graphite electrode is located. According to the temperature control method for the continuous graphitization furnace, the control strategy based on data driving and the intelligent model can exceed traditional simple feedback control, accurate prediction and active intervention on the temperature of the graphitization furnace are achieved, and therefore complex thermodynamic response caused by the fact that the characteristics of materials in the furnace change along with the temperature is effectively coped with; and the graphitization process is always in the optimal temperature interval.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of temperature control, in particular to a continuous graphitization furnace temperature control method and system. BACKGROUND

[0002] The graphitization furnace is a key equipment for high-temperature treatment of carbon materials. It heats up through graphite electrodes to make the materials in the furnace realize the ordered arrangement of carbon atoms at high temperature, and is widely used in new energy, metallurgy and other fields. The physical and chemical properties of graphite are largely determined by the graphitization temperature. The temperature for producing high-quality graphite products should reach 2800-3000℃, so the control of the heating temperature of the graphitization furnace is an important factor to determine the graphitization degree and product quality of graphite materials.

[0003] When carbon materials are graphitized, the process is to control the heating temperature at different stages. According to the temperature control stage range, the whole process can be divided into the following three stages: 1. Repeated baking stage (room temperature to 1250℃): The carbon material baked at about 1250℃ has initial thermal and impact properties, and the temperature rising speed should be accelerated to make the baked material complete the preheating transition stage as soon as possible. In this stage, the physical structure of the carbon material has little change.

[0004] 2. Strict temperature rising stage (1250℃ to 1800℃): This is the key temperature range for graphitization. In this range, the physical structure and chemical composition of the carbon material will change greatly, i.e. the disordered layer structure of amorphous carbon will have a tendency to change to a graphite crystal structure, while the unstable low molecular hydrocarbon and some impurity element groups on the edge of the amorphous carbon microcrystalline structure will continuously decompose and escape, thus changing the structure and causing defects. Therefore, the side effects of thermal stress and the concentration of the stress must be reduced, and the temperature rising rate in this stage must be strictly controlled.

[0005] 3. Free temperature rising stage (1800℃ to 3000℃ and above): In this temperature range, the transformation of the graphite crystal structure of the carbon material is basically completed, and uninterrupted heating can continuously increase the temperature to deepen the graphitization of the material. The important factor affecting the quality of the finished product after the graphitization of the carbon material is the highest temperature. Therefore, the heating temperature should be as high as possible.

[0006] The temperature control method of the traditional continuous graphitization furnace usually only relies on the real-time temperature value of a certain temperature measurement point in the furnace, and adjusts the power through the PID algorithm. However, the temperature control method of the traditional graphitization furnace is based on single temperature feedback closed-loop regulation, such as obtaining the temperature in the furnace through a thermocouple or an infrared sensor, and adjusting the electrode power combined with the PID algorithm. However, the internal temperature gradient of the material in the graphitization furnace and the heat capacity of the furnace body cause the temperature change rate (dT / dt) to lag behind the actual thermal state. Therefore, relying only on the real-time temperature cannot accurately predict the thermal imbalance trend, and cannot well control the temperature in the three stages of the graphitization furnace. SUMMARY

[0007] The purpose of the present application is to provide a continuous graphitization furnace temperature control method and system based on multi-dimensional parameters combined to accurately control the temperature in the graphitization furnace.

[0008] In order to achieve the above purpose, the present application provides a continuous graphitization furnace temperature control method, the graphitization furnace comprising a furnace body and a pair of graphite electrodes inserted in the furnace body, one end of the furnace body being provided with a feeding port, the other end of the furnace body being provided with a discharging port, the feeding port being connected to a distributor through a first valve, and the discharging port being connected to a cooling kiln through a second valve; the temperature control method comprising: real-time detection of the real-time temperature of the material in the furnace body and the instantaneous power of the circuit where the graphite electrodes are located; based on the real-time temperature, calculating the temperature change rate, and based on the instantaneous power, calculating the power change rate; processing the real-time temperature, the temperature change rate, and the power change rate based on an intelligent model configured by a neural network architecture to obtain a first adjustment value for adjusting the output power of the circuit where the graphite electrodes are located.

[0009] Preferably, the cumulative power is also calculated based on the instantaneous power, and the cumulative power is sent into the intelligent model together with the real-time temperature, the temperature change rate, and the power change rate.

[0010] Preferably, the intelligent model adopts a hybrid architecture of convolutional neural network and long short-term memory network.

[0011] Preferably, a plurality of partitions of the furnace body in the axial direction are respectively provided with infrared temperature sensors, and the real-time temperature is obtained by comprehensive calculation of the detection values of a plurality of infrared temperature sensors.

[0012] Preferably, the intelligent model also outputs a second adjustment value, which is used to adjust the state of the first valve and the second valve.

[0013] Preferably, the intelligent model is adapted to process parameters of different graphite raw materials through transfer learning, and initial training data is derived from historical production data of multiple groups of carbon fiber and petroleum coke mixtures.

[0014] Preferably, input and output data of the intelligent model are stored to obtain a historical data set, and the intelligent model is updated and trained based on the historical data set.

[0015] The application also provides a continuous graphitization furnace temperature control system, which comprises a furnace body and a pair of graphite electrodes inserted into the furnace body, one end of the furnace body is provided with a feeding port, the other end of the furnace body is provided with a discharging port, the feeding port is connected with a distributor through a first valve, and the discharging port is connected with a cooling kiln through a second valve; the temperature control system works based on the continuous graphitization furnace temperature control method as described above.

[0016] The application also provides a continuous graphitization furnace temperature control system, which comprises: one or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the programs comprise instructions for executing the continuous graphitization furnace temperature control method as described above.

[0017] The application also provides a computer-readable storage medium comprising a computer program, which can be executed by a processor to complete the continuous graphitization furnace temperature control method as described above.

[0018] Compared with the prior art, the continuous graphitization furnace temperature control method provided by the application can reflect the dynamic trend of the graphitization process by detecting the real-time temperature of the material in the furnace and the instantaneous power of the circuit where the graphite electrode is located in real time, and further calculating the temperature change rate and the power change rate based on these real-time data. Subsequently, these data are input into the pre-trained intelligent model as parameters for processing, and the intelligent model can output a first adjustment value, based on which the output power of the circuit where the graphite electrode is located is adjusted in real time. This control strategy based on data driving and intelligent model can overcome the simple feedback control of the traditional method, realize accurate prediction and active intervention of the temperature of the graphitization furnace, and effectively cope with the complex thermodynamic response caused by the change of the material properties in the furnace with temperature, so as to ensure that the graphitization process is always in the best temperature range. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 It is a continuous graphitization equipment principle diagram in the embodiment of the application.

[0020] Figure 2 It is a temperature control method flowchart in the embodiment of the application. DETAILED DESCRIPTION

[0021] To describe the technical content, structural features, achieved purposes and effects of the present application in detail, the following will be described in detail in combination with the embodiments and the accompanying drawings.

[0022] The embodiment discloses a continuous graphitization furnace temperature control method, which comprises the following steps: Figure 1 The graphitization furnace comprises a furnace body 1 and a pair of graphite electrodes J1 inserted in the furnace body 1, one end of the furnace body 1 is provided with a feeding port K1, the other end of the furnace body 1 is provided with a discharging port K2, the feeding port K1 is connected with a distributor 2 through a first valve F1, and the discharging port K2 is connected with a cooling kiln 3 through a second valve F2.

[0023] For this purpose, please refer to Figure 1 and Figure 2 The temperature control method in the embodiment comprises the following steps: S1: real-time detection of the real-time temperature of the material in the furnace body 1 and the instantaneous power of the circuit in which the graphite electrode J1 is located; S2: based on the real-time temperature, calculating the temperature change rate, and based on the instantaneous power, calculating the power change rate; S3: based on the intelligent model configured by the neural network architecture, processing the real-time temperature, the temperature change rate and the power change rate to obtain a first adjustment value for adjusting the output power of the circuit in which the graphite electrode J1 is located.

[0024] The present application introduces multi-dimensional parameters (real-time temperature, temperature change rate and power change rate) and combines an intelligent model based on a neural network architecture, aiming to solve the problems of insufficient temperature control accuracy, strong hysteresis and inability to perform fine stage control in the prior art.

[0025] The working principle is that the real-time temperature of the material in the furnace body 1 and the instantaneous power of the circuit in which the graphite electrode J1 is located are first collected in real time.

[0026] Subsequently, these real-time data are processed to calculate the temperature change rate (dT / dt) and the power change rate (dP / dt). These change rate parameters can effectively reflect the dynamic trend of the heat state in the furnace and the instantaneous change of the energy input, making up for the deficiency that a single real-time value cannot predict thermal imbalance.

[0027] Finally, these multi-dimensional parameters are sent to the pre-trained intelligent model. The intelligent model uses its powerful pattern recognition and nonlinear mapping capabilities to comprehensively analyze these input data, learns and understands the complex thermodynamic and kinetic laws in the graphitization process, and thus outputs a first adjustment value for accurately adjusting the output power of the circuit in which the graphite electrode J1 is located.

[0028] The aforementioned temperature control method, firstly, through real-time monitoring and calculation of the rate of change of temperature and power, can more accurately and earlier perceive the dynamic changes in the thermal state within the furnace and the instantaneous fluctuations in energy input, overcoming the hysteresis issues caused by heat capacity and temperature gradients in traditional methods. This enables the intelligent model to perform more precise temperature prediction and control, reducing the temperature fluctuation range within the furnace from the traditional ±50°C to within ±5°C, ensuring that the graphitization process proceeds in a more stable temperature environment.

[0029] Secondly, because the intelligent model can comprehensively analyze multi-dimensional data, it can achieve fine-grained adjustment of the output power of the graphite electrode J1, avoiding overshoot or undershoot, thereby reducing unnecessary energy loss. For example, during the strict control of the heating stage, by precisely controlling the heating rate, excessive energy consumption can be avoided, resulting in a 15%-20% increase in energy efficiency for the entire graphitization process.

[0030] Furthermore, precise temperature control is directly related to the degree of graphitization and the perfection of the crystal structure of the graphite material. The present invention can control the maximum temperature within a very narrow range of 2900°C ± 5°C, which is crucial for producing high-quality graphite products. It significantly improves product uniformity and key properties such as electrical and thermal conductivity.

[0031] Furthermore, the introduction of intelligent models enables it to handle complex nonlinear relationships, making it more robust to external interference and material variations. By analyzing the rate of change of temperature and power, the system can better identify and respond to abnormal situations, reducing downtime and improving the continuous operation of the production line.

[0032] It should be noted that the temperature and power rate of change can be calculated in a variety of ways. In addition to simple differential calculations, algorithms such as Kalman filtering, sliding average, or exponentially weighted moving average can be used to smooth the data and calculate the rate of change. This can reduce the impact of noise on the calculation results and improve the accuracy and stability of the rate of change.

[0033] On the other hand, the above temperature control method is further improved: the cumulative power is calculated based on the instantaneous power, and the cumulative power is sent to the intelligent model together with the real-time temperature, temperature change rate, and power change rate.

[0034] In this embodiment, while measuring instantaneous power in real time, it is also integrated over time to obtain the cumulative power within a certain time window. This cumulative power reflects the total energy absorbed by the furnace body 1 and the material, providing macroscopic information about the thermal inertia and overall energy accumulation within the furnace.

[0035] When the cumulative power is sent into the intelligent model together with the real-time temperature, temperature change rate, and power change rate, the intelligent model can not only capture the instantaneous and dynamic thermal state, but also combine the "thermal history" of the material to more comprehensively understand the thermodynamic process in the furnace. For example, even if the real-time temperature and instantaneous power seem normal, but if the cumulative power is too high or too low, it may indicate that the graphitization degree of the material will deviate from the expected value. Through such multi-dimensional input containing historical cumulative information, the intelligent model can more accurately predict the final thermal state and graphitization degree of the material in the furnace, thereby outputting more accurate first adjustment values and achieving more stable and process requirement conforming temperature control.

[0036] The calculation of cumulative power can be deformed in various ways. For example, different cumulative time windows can be set according to different process requirements, such as calculating the cumulative power of the past 1 hour, 4 hours, or the entire graphitization process. In addition, besides simple accumulation, weighted average or exponential decay can also be introduced to calculate the cumulative power, so that the recent power has a greater impact on the cumulative value, thereby better reflecting the current energy accumulation trend.

[0037] In addition to receiving cumulative power as a direct input, the intelligent model can also further process the cumulative power through feature engineering, such as calculating the change rate of cumulative power, the ratio of cumulative power to real-time temperature, and other derived features, to mine more potential correlation information and use it as the input of the model, thereby further improving the prediction ability of the model.

[0038] On the other hand, the intelligent model adopts a hybrid architecture of convolutional neural network (CNN) and long short-term memory network (LSTM). The CNN is used to extract the spatio-temporal features of temperature and power, and the LSTM is used to capture the dynamic change law of time series.

[0039] CNN performs well in processing data with local correlation. In the scenario of graphite furnace temperature control, temperature and power data can be regarded as "images" or "signals" with spatio-temporal characteristics. For example, the temperature sequence within a period of time can be regarded as a one-dimensional signal, while the temperature data of multiple distributed sensors can form a spatial correlation. CNN slides on these data through its convolution kernel, automatically extracting local features such as instantaneous fluctuations of temperature, sudden changes in power, or spatial features of temperature gradient between different sensors. These "spatio-temporal features" can be short-term trends, abnormal points, or specific patterns, which are crucial for judging the local thermal state and instantaneous energy response in the furnace.

[0040] LSTM is a special type of recurrent neural network (RNN) designed to address the gradient vanishing or exploding problem that traditional RNNs face when dealing with long sequence data, enabling it to effectively capture long-term dependencies in time series. In the graphitization process, the change in furnace temperature is not only influenced by the current power input, but also closely related to the heating history over a longer period of time, material characteristics, heat accumulation, and other factors. LSTM, through its unique "gating" mechanism (input gate, forget gate, output gate), can selectively remember or forget historical information, effectively capturing the dynamic changes in the temperature and power time series, such as the long-term trend of the heating rate, the time required to reach a certain temperature, and the transition characteristics between different stages.

[0041] In this embodiment, by combining CNN and LSTM, the intelligent model first uses CNN to extract rich spatio-temporal features from the original temperature and power data, which can be local and high-dimensional representations. Subsequently, these features extracted by CNN are input into the LSTM network. LSTM further captures the dynamic rules and long-term dependencies of the evolution of these feature sequences over time based on these feature sequences. This hybrid architecture allows the intelligent model to balance both local details and overall temporal trends in the data, providing a more comprehensive and in-depth understanding and modeling of the complex heating process in the graphitization furnace, ultimately outputting more accurate first adjustment values and achieving fine-grained and adaptive control of the temperature in the graphitization furnace. Especially in the strict control of the heating stage, it can better predict and respond to temperature changes, ensuring product quality.

[0042] On the other hand, the multiple zones in the axial direction of the furnace body 1 are respectively provided with infrared temperature sensors T, and the real-time temperature is obtained by comprehensive calculation of the detection values of the infrared temperature sensors T.

[0043] The infrared temperature sensors T are distributed in the multiple zones in the axial direction of the furnace body 1, which means that there are temperature monitoring points at different positions of the furnace body 1 along the material flow direction. This layout can overcome the limitations of a single temperature measurement point and provide temperature profile information along the axial direction of the furnace, thus more comprehensively reflecting the actual temperature of the material at different heating stages.

[0044] Here, "comprehensive calculation" includes simple arithmetic mean, and can also be: Weighted average: different weights are given according to the importance of different zones or the reliability of infrared temperature sensors T. For example, in the key area of strict temperature control, the weight of the infrared temperature sensor T may be higher.

[0045] Zone average: the infrared temperature sensor T data in each zone is averaged, and then the average values of each zone are further processed.

[0046] Spatial Interpolation: Utilize the data from multiple infrared temperature sensors T, combined with the geometric structure of the furnace body 1 and the heat conduction model, estimate the entire axial or even radial temperature distribution through interpolation algorithms (such as Kriging interpolation, inverse distance weighting, etc.), to obtain a more representative "real-time temperature" value.

[0047] Model-based Fusion: Input the infrared temperature sensor T data into a simplified thermodynamic model, combined with material flow, power input, etc. information, output a comprehensive real-time temperature through the model.

[0048] Statistical Analysis: Calculate the variance, standard deviation, etc. of the temperature to evaluate the temperature uniformity.

[0049] On the other hand, the intelligent model also outputs a second adjustment value, which is used to adjust the state of the first valve F1 and the second valve F2.

[0050] The present application not only outputs the first adjustment value for adjusting the power of the graphite electrode J1, but also outputs the second adjustment value for adjusting the state of the first valve F1 and the second valve F2, thereby realizing the coordinated control of the graphitization furnace.

[0051] The working principle is that after receiving and processing the multi-dimensional input data of real-time temperature, temperature change rate, power change rate and cumulative power, the intelligent model not only understands the thermodynamic state and energy input situation in the furnace, but also further analyzes the potential correlation between these information and material flow. For example, if the intelligent model judges that the temperature in the furnace has a downward trend, in addition to increasing the heating power, it may also suggest slowing down the feeding speed (by adjusting the first valve F1) to reduce the impact of cold material on the temperature in the furnace; or if the model judges that the temperature in the furnace is too high and the material has completed graphitization, it may suggest speeding up the discharging speed (by adjusting the second valve F2) to quickly remove the high-temperature material.

[0052] In addition, the second adjustment value can be a continuous analog quantity for precise control of the opening of the valve, thereby realizing continuous adjustment of the material flow. For example, 0-100% opening proportional control.

[0053] The second adjustment value can also be a discrete instruction, such as "full open", "full close", "half open", etc., suitable for scenes that only require rough control of material import and export.

[0054] This multi-variable coordinated optimization can more finely manage the heat balance in the furnace, for example, when the temperature in the furnace decreases, in addition to increasing the power, it can also slow down the feeding speed simultaneously, thereby more effectively stabilizing the furnace temperature and further reducing the temperature fluctuation range to ±2℃, significantly improving the accuracy and response speed of temperature control.

[0055] On the other hand, the intelligent model adapts the process parameters of different graphite raw materials through transfer learning. The initial training data comes from historical production data of multiple groups of carbon fiber and petroleum coke mixtures.

[0056] Specifically, the initial training data of the intelligent model comes from historical production data of multiple groups of carbon fiber and petroleum coke mixtures. These historical data contain information such as temperature, power, material in-out speed, and final product quality of carbon fiber and petroleum coke mixtures in the graphitization furnace under different proportions, different batches, and different process conditions.

[0057] By pre-training on these large-scale and diverse data sets, the intelligent model can learn general rules and basic features in the graphitization process, such as: the response mode of the material to power input in different temperature intervals, the phase change characteristics of the material under different heating rates, and the typical behavior of various raw material mixtures at different stages. This pre-training stage enables the model to have a preliminary understanding and generalization ability of the graphitization process.

[0058] When the intelligent model needs to be used for new graphite raw materials that are not fully covered in the initial training data (for example, a new type of carbon-based material or a special proportion mixture), instead of starting from scratch, the pre-trained intelligent model is used as the base model. Through transfer learning, the model is fine-tuned or feature extracted on a small amount of historical production data of the new raw material.

[0059] Fine-tuning: usually freezes the bottom network layers of the pre-trained model that learn general features (e.g., early layers of CNN or certain units of LSTM), and only trains the high layers (e.g., fully connected layers before the output layer) or all layers with a small learning rate. This allows the model to quickly and efficiently learn the specific process parameters and behavior patterns of the new raw material while retaining general knowledge.

[0060] Feature extraction: use the pre-trained model as a feature extractor, pass the input data of the new raw material through the model, extract high-dimensional feature representations, and then input these features into a new, simpler classifier or regressor for training.

[0061] Through transfer learning, the intelligent model can "transfer" the general knowledge learned from a large amount of historical data to new graphite raw materials, enabling it to quickly and accurately adapt to the process parameters of new raw materials and output the first adjustment value and optional second adjustment value for adjusting the output power of the graphite electrode J1, ensuring the performance and stability of the temperature control system when processing diverse raw materials.

[0062] In another aspect, the operation of the graphitization furnace is a long-term and dynamic process. Over time, factors such as the aging of the furnace body 1, the wear of the graphite electrodes J1, the slight differences in batches of materials, and changes in environmental conditions can cause slow drifts in the characteristics of the graphitization process, causing the performance of the initially trained intelligent model to gradually decline. Traditional static models cannot maintain optimal control effects when facing such "concept drift", which can lead to decreased temperature control accuracy, increased energy consumption, or fluctuations in product quality. Therefore, a mechanism is needed to enable the intelligent model to continuously learn and adapt to these changes. To this end, the temperature control method in the present embodiment further comprises: storing the input and output data of the intelligent model to obtain a historical data set, and updating and training the intelligent model based on the historical data set.

[0063] In the daily operation of the graphitization furnace, each input of the intelligent model (including real-time temperature, temperature change rate, power change rate, cumulative power, etc.) and its corresponding output (first adjustment value, second adjustment value) are recorded and stored in real time, forming a growing historical data set. This historical data set contains the "experience" and "performance" of the intelligent model in the actual production environment, reflecting the model's decisions and their corresponding actual effects under different working conditions.

[0064] Through this continuous data accumulation and model updating and training mechanism, the intelligent model can continuously learn from new production data, capture subtle changes in process characteristics, and correct its own prediction and control strategies. This enables the intelligent model to continuously adapt to factors such as the aging of the furnace body 1, the slight changes in material characteristics, and changes in environmental conditions, thereby maintaining optimal control performance at all times and ensuring the long-term stability of the graphitization process and continuous improvement of product quality.

[0065] In another preferred embodiment of the present application, a continuous graphitization furnace temperature control system is also disclosed, which includes a furnace body 1 and a pair of graphite electrodes J1 inserted in the furnace body 1. One end of the furnace body 1 is provided with a feed inlet K1, and the other end of the furnace body 1 is provided with a discharge outlet K2. The feed inlet K1 is connected to the distributor 2 through the first valve F1, and the discharge outlet K2 is connected to the cooling kiln 3 through the second valve F2. The temperature control system works based on the continuous graphitization furnace temperature control method in the above embodiment.

[0066] The application also discloses another temperature control system, which comprises one or more processors, a memory and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the programs comprise instructions for executing the temperature control method as described above. The processor can be a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC) or one or more integrated circuits, which are used to execute the related programs to realize the functions required by the modules in the temperature control system of the embodiments of the application or execute the temperature control method of the method embodiments of the application.

[0067] The application also discloses a computer readable storage medium comprising a computer program, which can be executed by a processor to complete the temperature control method as described above. The computer readable storage medium can be any available medium accessible by a computer or a data storage device such as a server, data center and the like integrated with one or more available media. The available medium can be a read-only memory (ROM), a random access memory (RAM), a magnetic medium such as a floppy disk, a hard disk, a magnetic tape, a magnetic disc or an optical medium such as a digital versatile disc (DVD), or a semiconductor medium such as a solid state disk (SSD) and the like.

[0068] The embodiments of the application also disclose a computer program product or a computer program, which comprises computer instructions stored in a computer readable storage medium. A processor of an electronic device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions to enable the electronic device to execute the temperature control method.

[0069] The above only discloses the preferred embodiments of the application, and of course cannot limit the scope of the application, so the equivalent changes made in the patent scope of the application still fall within the scope of the application.

Claims

1. A continuous graphitization furnace temperature control method, characterized in that: The graphitization furnace includes a furnace body and a pair of graphite electrodes inserted in the furnace body. A feed port is provided at one end of the furnace body, and a discharge port is provided at the other end of the furnace body. The feed port is connected to a distributor through a first valve, and the discharge port is connected to a cooling kiln through a second valve. The temperature control method includes: Real-time detection of the real-time temperature of the material in the furnace body and the instantaneous power of the circuit where the graphite electrode is located; Calculating a temperature change rate based on the real-time temperature, and calculating a power change rate based on the instantaneous power; An intelligent model configured based on a neural network architecture processes the real-time temperature, the temperature change rate, and the power change rate to obtain a first adjustment value for adjusting the output power of the circuit where the graphite electrode is located.

2. The temperature control method of a continuous graphitization furnace according to claim 1, characterized in that: The accumulated power is also calculated based on the instantaneous power, and the accumulated power is sent to the intelligent model together with the real-time temperature, the temperature change rate, and the power change rate.

3. The temperature control method of a continuous graphitization furnace according to claim 1, characterized in that: The intelligent model adopts a hybrid architecture of convolutional neural network and long short-term memory network.

4. The temperature control method of a continuous graphitization furnace according to claim 1, characterized in that: Infrared temperature sensors are respectively provided in the plurality of axial partitions of the furnace body, and the real-time temperature is obtained by comprehensive calculation of the detection values ​​of a plurality of the infrared temperature sensors.

5. The temperature control method of a continuous graphitization furnace according to claim 1, characterized in that: The intelligent model further outputs a second adjustment value, where the second adjustment value is used to adjust the states of the first valve and the second valve.

6. The temperature control method of a continuous graphitization furnace according to claim 1, characterized in that: The intelligent model adapts to the process parameters of different graphite raw materials through transfer learning, and the initial training data comes from historical production data of multiple sets of carbon fiber and petroleum coke mixtures.

7. The temperature control method of a continuous graphitization furnace according to claim 1, characterized in that: The input and output data of the intelligent model are stored to obtain a historical data set, and the intelligent model is updated and trained based on the historical data set.

8. A continuous graphitization furnace temperature control system, characterized in that: The graphitization furnace includes a furnace body and a pair of graphite electrodes inserted in the furnace body. A feed port is provided at one end of the furnace body, and a discharge port is provided at the other end of the furnace body. The feed port is connected to a distributor through a first valve, and the discharge port is connected to a cooling kiln through a second valve. The temperature control system operates based on the continuous graphitization furnace temperature control method according to any one of claims 1 to 7.

9. A continuous graphitization furnace temperature control system, characterized in that: include: one or more processors; Memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the programs including instructions for executing the continuous graphitization furnace temperature control method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The method comprises a computer program, which can be executed by a processor to implement the temperature control method of a continuous graphitization furnace according to any one of claims 1 to 7.

Images