Temperature uniformity optimization treatment system of graphitization furnace

Abstract

The invention relates to the field of intelligent temperature control of industrial kilns, and discloses a graphitization furnace temperature uniformity optimization processing system which comprises a temperature monitoring module, a dynamic matrix control module, a material-control coupling module, a lattice feedback module and a multi-objective optimization module. The method comprises the following steps: constructing an equivalent temperature field through multi-source temperature data fusion; predicting a graphitization trend and generating a dynamic matrix control instruction in combination with a multi-physical field model; dynamically adjusting a thermal conductivity matching temperature gradient of the guard plate; online correcting a material constant based on lattice feedback; closed-loop control is formed, and the process stability and energy efficiency of the graphitization furnace are improved. Through multi-source fusion monitoring, multi-physical field coupling modeling, dynamic thermal conductivity regulation and control, lattice feedback online correction and reinforcement learning multi-objective optimization, temperature uniformity control and energy efficiency improvement of the graphitization furnace are realized, and material performance and process economy are guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of intelligent temperature control of industrial furnaces, and particularly to a temperature uniformity optimization processing system for a graphitization furnace. Background Art

[0002] With the surging demand for highly conductive graphite materials in the new energy industry, the temperature field uniformity of the graphitization smelting process has become the core factor restricting the crystal orientation degree and electrical conductivity of materials. As a key production equipment, the stability of the thermal field distribution of the graphitization furnace directly affects the microstructure orderliness and batch consistency of materials. However, the existing technologies still face significant bottlenecks in continuous production with high efficiency and low energy consumption: Traditional graphitization furnaces generally adopt a fixed insulation guard plate and a single-point thermocouple monitoring mode. Due to the unreasonable distribution of the thermal resistance of the insulation system in the furnace head area, the axial temperature gradient changes abnormally steeply, resulting in local overheating or under-temperature phenomena. The thermal radiation loss of the furnace lining material and the non-adjustable thermal conductivity characteristics of the guard plate in the high-temperature environment further exacerbate the non-uniformity of the thermal field distribution, causing energy waste and material property fluctuations. Although the industry has tried to introduce silicon carbide-based composite linings and zone temperature control strategies, limited by the static insulation structure design and the insufficient spatial resolution of discrete temperature measurement points, it is still difficult to achieve precise control of the temperature field in the entire furnace chamber.

[0003] The current temperature control systems mostly rely on offline optimization models driven by historical data and lack the dynamic response ability to the graphitization process. The mismatch between the carbon atom migration kinetic parameters and the real-time process conditions leads to a decrease in the adaptability of the temperature prediction model during the material phase change stage. In addition, the thermal physical properties of traditional guard plate materials are fixed and cannot actively adjust the heat flow direction according to the temperature gradient change, resulting in hysteresis and limitations in temperature uniformity control.

[0004] Therefore, the present invention proposes a temperature uniformity optimization processing system for a graphitization furnace to solve the deficiencies of the existing technologies. Summary of the Invention

[0005] Aiming at the deficiencies of the existing technologies, the present invention provides a temperature uniformity optimization processing system for a graphitization furnace, which solves the problems of poor electrical conductivity, low crystal orientation degree, and high energy consumption of graphite materials caused by uneven temperature field distribution, lag in thermal conductivity regulation, and material-temperature control decoupling in traditional graphitization furnaces.

[0006] To achieve the above objectives, the present invention is realized through the following technical solutions: A temperature uniformity optimization processing system for a graphitization furnace, the system includes: A temperature monitoring module, including an infrared thermal imager array and a distributed thermocouple group. The infrared thermal imager array collects the two-dimensional temperature field distribution in the furnace in multiple spectra, and the thermocouple group is distributed axially and circumferentially symmetrically in the furnace insulation layer to output multi-source temperature data; a dynamic matrix control module that receives the multi-source temperature data in real time, inputs the two-dimensional temperature field distribution data into a multi-physics field model that combines the temperature field transient heat transfer equation and the carbon atom migration kinetics equation, and generates heating power control instructions for each temperature zone through calculation using the multi-physics field model; a material-control coupling module comprising a ZrO2-SiC gradient guard plate and an electromagnetic control unit, wherein the electromagnetic control unit regulates the electron concentration of the interface phase of the gradient guard plate by applying an alternating electromagnetic field according to the control instruction, and dynamically adjusts the axial thermal conductivity distribution of the guard plate to match the real-time temperature gradient distribution obtained by the temperature monitoring module; A lattice feedback module includes a high-temperature resistant XRD probe that penetrates the guard plate layer and a diffraction spectrum analysis unit. The XRD probe transmits X-rays through the graphitized material and receives diffraction signals to obtain crystal plane diffraction data in real time. The diffraction spectrum analysis unit fits the diffraction peak intensity using a Lorentz function to generate a graphitization degree feedback signal and transmits it to the multi-physics field model to correct the material constants in the carbon atom migration kinetics equation; The multi-objective optimization module receives the control instructions and the graphitization degree feedback signal, dynamically optimizes the temperature uniformity, crystal orientation and unit energy consumption indicators through a reinforcement learning algorithm, generates a process parameter set and feeds it back to the dynamic matrix control module to form a closed-loop control.

[0007] Preferably, in the temperature monitoring module: The infrared thermal imager array collects two-dimensional temperature field distribution data in the furnace using multi-spectral bands ,in, Indicates the horizontal coordinate; represents the axial coordinate; represents the time variable; The distributed thermocouple group is divided into Segmental and circumferential The zones are evenly arranged, and the total number of temperature measurement points is , output temperature data of each temperature zone ; The fusion of multi-source temperature data calculates the equivalent temperature of each temperature zone through the following formula: ; in, For the Infrared image pixel area corresponding to the temperature zone; for The total number of pixels in the region; and is the fusion weight coefficient, satisfying .

[0008] Preferably, in the dynamic matrix control module: The multi-physical field model consists of the following simultaneous equations: Transient heat transfer equation of the temperature field: ; where, is the axial temperature distribution in the furnace; is the equivalent thermal diffusivity; is the heating power-temperature conversion coefficient; is the axial heating power density distribution; is the influence coefficient of the carbon source mass flow rate; is the carbon source feeding rate; Kinetics equation of carbon atom migration: ; where, is the graphitization degree; is the axial temperature standard deviation; is the material kinetics constant.

[0009] Preferably, the dynamic matrix control module further includes: Predictive control algorithm: Based on the step response matrix , construct the prediction time domain and the control time domain , and the objective function is: ; where, is the target temperature distribution; is the temperature tracking weight matrix, and the diagonal element ; is the control increment weight matrix, and the diagonal element ; is the heating power increment vector at the th moment; Solution rule: Solve the optimal control sequence through the quadratic programming algorithm, satisfying: ; where , are the power increment constraints.

[0010] Preferably, the material-control coupling module includes: Gradient baffle thermal conductivity regulation equation: The axial thermal conductivity distribution of the baffle is dynamically adjusted according to the real-time temperature gradient, satisfying: ; where, is the axial thermal conductivity distribution of the baffle; is the matrix thermal conductivity; is the thermal conductivity control range; is a sign function, taking +1 when the temperature gradient is positive and -1 otherwise; Working rules of electromagnetic control unit: Frequency of applied alternating electromagnetic field , magnetic field strength ; The electromagnetic field changes the electron concentration of the ZrO2-SiC interface phase , adjust the thermal conductivity to meet: .

[0011] Preferably, the gradient guard plate structure: The ZrO2-SiC gradient guard plate is made of Layer composition, layer ZrO2 volume fraction satisfy: ; in, is the upper limit of ZrO2 volume fraction at the inlet end of the guard plate; is the lower limit of the outlet; Adjacent layers are connected by a Cr3C2 / NiCr interface layer, and the interface thickness is controlled within a preset range to reduce interlayer thermal stress and enhance mechanical bonding strength; Electromagnetic control optimization rules: Establishing thermal conductivity control range Mapping relationship with electromagnetic field parameters: ; in, is the material permeability constant; is the reference frequency; is the exponential coefficient.

[0012] Preferably, in the lattice feedback module: The high temperature resistant XRD probe comprises a tantalum carbide shield and a liquid metal dynamic sealing layer, which is used to transmit and receive X-ray signals in a high temperature environment, wherein the X-ray wavelength is The diffraction spectrum analysis unit fits the crystal plane diffraction peak intensity by Lorentz function , and the degree of graphitization is calculated according to the following formula : ; in, The integrated intensity of the whole diffraction spectrum is , and the thickness of the tantalum carbide shield satisfies the X-ray transmittance The thermal expansion coefficient of the liquid metal dynamic sealing layer matches the material of the protective cover.

[0013] Preferably, the liquid metal dynamic sealing layer is composed of a gallium-based alloy, the composition of which satisfies , at the X-ray wavelength Absorption coefficient under , and the thickness of the sealing layer is maintained by a circulation pump Constant flow rate Adaptive adjustment according to the temperature in the furnace.

[0014] Preferably, in the multi-objective optimization module: The optimization objective function is defined as: ; in, is the integrated intensity of the diffraction peak of the crystal plane, which is calculated by the diffraction peak intensity ratio output by the lattice feedback module; is the standard deviation of the axial temperature in the furnace, which is obtained in real time through the fusion data of the temperature monitoring module; is the energy consumption per unit mass, calculated based on the integral value of heating power and the amount of carbon source processed; is the dynamic weight coefficient, satisfying the normalization condition ; The weight coefficient is updated online through the reinforcement learning algorithm, and the update rule is: ; in, is the learning rate; The preset comprehensive benefit target value; It is a real-time comprehensive benefit indicator.

[0015] The present invention also provides a method for optimizing the temperature uniformity of a graphitization furnace, the method comprising the following steps: Step S1: acquiring two-dimensional temperature field distribution data in the furnace through an infrared thermal imager array, and simultaneously collecting local temperature data through an axial gradient-distributed thermocouple group, and outputting a multi-source temperature signal; Step S2: performing weighted fusion on the two-dimensional temperature field distribution data collected by the infrared thermal imager and the local temperature data of the thermocouple group according to the temperature zone division to generate equivalent temperature data of each temperature zone; Step S3: inputting the equivalent temperature data into a multi-physics field model of the simultaneous temperature field transient heat transfer equation and the carbon atom migration kinetics equation to predict the temperature field distribution and graphitization degree evolution trend in the future time domain; Step S4: Based on the prediction results of the multi-physics field model, the optimal adjustment amount of the heating power of each temperature zone is calculated by a dynamic matrix control algorithm to generate a heating power control instruction; Step S5: According to the heating power control instruction, the electron concentration of the interface phase of the ZrO2-SiC gradient guard plate is adjusted by an alternating electromagnetic field, and the axial thermal conductivity distribution of the guard plate is dynamically adjusted to match the temperature gradient direction monitored in real time; Step S6: obtaining the crystal plane diffraction intensity data of the graphite material in real time through a high temperature resistant X-ray diffraction probe, analyzing and calculating the current degree of graphitization, and feeding the data back into the carbon atom migration kinetic equation to correct the material kinetic constants online; Step S7: Combining the heating power control instructions, real-time graphitization degree and energy consumption data, a reinforcement learning algorithm is used to dynamically balance the temperature uniformity, crystal orientation and unit energy consumption indicators, generate an optimized set of process parameters and feed it back to step S4 to form a closed-loop control.

[0016] The present invention provides a temperature uniformity optimization processing system for a graphitization furnace. It has the following beneficial effects: 1. This invention solves the problem of large measurement errors of a single sensor in high-temperature and dusty environments through the collaborative data acquisition of an infrared thermal imager array and a distributed thermocouple group, combined with a spatial domain weighted fusion algorithm. It significantly improves the global coverage and local accuracy of furnace temperature field monitoring, and provides highly reliable input for subsequent control decisions.

[0017] 2. The present invention adopts a multi-physics field model that combines the transient heat transfer equation of the temperature field and the carbon atom migration kinetics equation. For the first time, it dynamically links the heat conduction process with the graphitization evolution of the material, breaking through the limitation of the traditional temperature control model that only considers the thermal field distribution. It realizes the coordinated prediction of the temperature field and material properties, and provides a multi-dimensional optimization basis for heating power regulation.

[0018] 3. The present invention dynamically adjusts the thermal conductivity distribution of the ZrO2-SiC gradient guard plate through an alternating electromagnetic field, so that the thermal properties of the guard plate match the temperature gradient changes in real time, solving the problem that traditional passive insulation structures cannot adapt to dynamic process conditions, significantly suppressing abnormal temperature fluctuations in the furnace, and ensuring the uniformity of the axial temperature field.

[0019] 4. The present invention uses real-time graphitization degree monitoring data from a high-temperature resistant XRD probe to online correct the material constants in the carbon atom migration kinetics equation, overcoming the prediction deviation problem of the traditional offline calibration model caused by material batch differences. This enables the multi-physics field model to have adaptability under dynamic working conditions and improves long-term control stability.

[0020] 5. The present invention dynamically adjusts the weight distribution of temperature uniformity, crystal orientation and energy consumption indicators through a reinforcement learning algorithm, constructs a closed loop of process parameter self-optimization, and solves the pain point of fixed weight strategy in controlling target conflicts under complex working conditions. While ensuring product quality, it reduces unit energy consumption and improves process economy. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 This is the system architecture diagram of the present invention; Figure 2 This is the method flow chart of the present invention. Specific embodiments

[0022] Next, in conjunction with the accompanying drawings of the present invention specification, the technical solutions in the embodiments of the present invention will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts shall fall within the protection scope of the present invention.

[0023] Please refer to Figure 1 , the embodiment of the present invention provides an optimization processing system for the temperature uniformity of a graphitization furnace, and the system includes: A temperature monitoring module, including an infrared thermal imager array and a distributed thermocouple group. The infrared thermal imager array acquires the two-dimensional temperature field distribution in the furnace in multiple spectra, and the thermocouple group is axially gradient and circumferentially symmetrically distributed in the furnace insulation layer to output multi-source temperature data; In this embodiment, the temperature monitoring module is composed of an infrared thermal imager array and a distributed thermocouple group, and is used to obtain and fuse the multi-dimensional temperature data in the furnace in real time. The specific implementation method is as follows: The infrared thermal imager array acquires the two-dimensional temperature field distribution data in the furnace chamber in multiple spectral bands , where represents the transverse coordinate; represents the axial coordinate; represents the time variable. The infrared thermal imager array is evenly arranged along the circumferential direction of the furnace body, covering the entire cross-section of the furnace chamber, and synchronously acquires the radiation energy signals at different wavelengths through multi-spectral detection technology, and inversely calculates the temperature field distribution in combination with Planck's radiation law. Preferably, the working band of the infrared thermal imager covers 8-14 μm to match the high-temperature radiation characteristics of the graphitization furnace, and the interference of carbon powder suspensions in the furnace on the optical path is eliminated through pixel-level temperature calibration.

[0024] The distributed thermocouple group is embedded in the furnace insulation layer according to the principle of axial gradient and circumferential symmetry. The specific arrangement method is as follows: The furnace body is axially equally spaced into segments, and each segment is evenly arranged with temperature measurement points in the circumferential direction, and the total number of temperature measurement points is . Preferably, the thermocouple adopts a K-type armored structure, the measurement end is embedded at a preset depth below the surface of the insulation layer, and the carbon powder pollution in the furnace is isolated through an alumina ceramic sleeve to output the local temperature data of each temperature zone .

[0025] The multi-source temperature data fusion is achieved through a spatial domain weighted algorithm, which specifically includes the following steps: Temperature zone division and pixel mapping: convert the two-dimensional temperature field distribution data collected by the infrared thermal imager into According to the axial and circumferential physical positions of the furnace body, it can be divided into temperature zones, each corresponding to a pixel area in the infrared image ; Spatial average calculation: for each temperature zone The infrared temperature data in the space is taken as the spatial average value, and the calculation formula is: ; in, for The total number of pixels in the region; Weighted fusion output: linearly weighted fusion is performed on the average infrared temperature and the thermocouple measurement value of the corresponding temperature zone to generate equivalent temperature data for each temperature zone: ; in, and is the fusion weight coefficient, satisfying Preferably, the weight coefficient is calibrated according to the measurement error distribution characteristics of the infrared thermal imager and the thermocouple, wherein, Value greater than , to strengthen the dominant role of two-dimensional temperature field distribution data.

[0026] The infrared camera array and distributed thermocouple group achieve data acquisition timing alignment through a time synchronization unit, with synchronization errors controlled within 1ms, ensuring spatiotemporal consistency of multi-source temperature data. The fusion weight coefficients are stored in the control system's non-volatile memory, supporting online recalibration based on furnace aging or sensor performance degradation.

[0027] a dynamic matrix control module that receives the multi-source temperature data in real time, inputs the two-dimensional temperature field distribution data into a multi-physics field model that combines the temperature field transient heat transfer equation and the carbon atom migration kinetics equation, and generates heating power control instructions for each temperature zone through calculation using the multi-physics field model; In this embodiment, the dynamic matrix control module generates heating power control instructions based on multi-physics field coupling modeling and predictive control algorithm. The specific implementation method is as follows: The multi-physics model is constructed by combining the temperature field transient heat transfer equation and the carbon atom migration kinetics equation to describe the dynamic coupling relationship between the temperature distribution in the furnace and the graphitization degree of the material. The temperature field transient heat transfer equation is defined as: ; in, represents the axial temperature distribution in the furnace, and its physical meaning is the temperature field along the axial position of the furnace body at time ; is the equivalent thermal diffusivity, which is obtained by comprehensively calculating the thermal conductivity, specific heat capacity and density of the furnace body material; is the heating power-temperature conversion coefficient; represents the axial heating power density distribution, which is generated by converting the current signals of the heating elements in each temperature zone; is the carbon source mass flow rate influence coefficient; it reflects the feeding rate on the perturbation effect of the temperature field, and is correlated with the process parameters through experimental calibration.

[0028] The carbon atom migration kinetic equation is defined as: ; wherein, represents the graphitization degree based on the (002) crystal plane of graphite, and its value range is from 0 to 1, and 1 represents complete graphitization; is the axial temperature standard deviation, which is obtained by calculating the real-time temperature data and is used to quantify the temperature uniformity; is the material kinetic constant, which is related to the carbon source type and the furnace atmosphere and is determined by off-line experimental fitting.

[0029] The predictive control algorithm is implemented based on the dynamic matrix control (DMC) framework, and specifically includes the following steps: Step response matrix construction: By applying a step power signal to the heating elements in each temperature zone, recording the temperature response curves of each temperature zone, and extracting the steady-state gain and dynamic characteristic parameters, construct dimensional dynamic matrix , wherein, is the total number of temperature zones, and the matrix element characterizes the influence coefficient of the power change in the temperature zone on the temperature in the temperature zone; Rolling optimization objective function design: ; wherein, is the target temperature distribution; is the temperature tracking weight matrix, and its diagonal element , which is designed by normalizing the temperature standard deviation of each temperature zone; is the control increment weight matrix, and its diagonal element element , which is used to constrain the adjustment range of the heating power; is the heating power increment vector at the th moment; Constraints and solution: Solve the optimal control sequence through the quadratic programming algorithm , satisfying the upper and lower limits of power increment constraints , where the constraint values are set according to the rated power and safety margin of the heating element.

[0030] Preferably, the update period of the dynamic matrix matches the thermal inertia time constant of the furnace body to avoid control deviation caused by model mismatch. The parameters of the weight matrix and are dynamically adjusted according to the process priority. For example, during the stage of rapid change in graphitization degree, increase the weight to enhance the temperature tracking accuracy.

[0031] Material-control coupling module, including ZrO2-SiC gradient shield and electromagnetic regulation unit. The electromagnetic regulation unit adjusts the interfacial phase electron concentration of the gradient shield according to the regulation instruction, and dynamically adjusts the axial thermal conductivity distribution of the shield to match the real-time temperature gradient distribution obtained by the temperature monitoring module; In this embodiment, the material-control coupling module realizes the active matching of the temperature gradient distribution in the furnace through the synergistic action of the dynamic regulation of the gradient shield thermal conductivity and the electromagnetic field. The specific implementation method is as follows: The ZrO2-SiC gradient shield adopts a multi-layer composite structure design and is stacked along the axial direction of the furnace body layers of functionally graded materials. The layer has a decreasing ZrO2 volume fraction along the axial direction, and its distribution law satisfies: ; where, is the upper limit of the ZrO2 volume fraction at the inlet end of the shield; is the lower limit at the outlet end. Preferably, adjacent layers are connected by a Cr3C2 / NiCr interface layer, and the interface thickness is controlled within a preset range to reduce the interlayer thermal stress and enhance the mechanical bonding strength. The interface thickness is .

[0032] The thermal conductivity distribution of the gradient shield is dynamically adjusted according to the direction of the real-time temperature gradient. The regulation equation is defined as: ; where, is the axial thermal conductivity distribution of the shield; is the thermal conductivity of the matrix material, which is calculated by weighting the thermal conductivities of the ZrO2 and SiC phases according to the volume fraction; is the thermal conductivity regulation amplitude; is a sign function that takes +1 when the temperature gradient direction is consistent with the preset positive direction and -1 otherwise. The introduction of the sign function enables the thermal conductivity of the guard plate to respond quickly when the temperature gradient direction changes abruptly, enhancing the compensation ability for local hot spots or cold regions.

[0033] The electromagnetic regulation unit adjusts the electron concentration of the guard plate interface phase by applying an alternating electromagnetic field, thereby changing the thermal conductivity regulation amplitude. . The electromagnetic field parameters include frequency and magnetic field strength , and its working rules satisfy: ; wherein, is the electron concentration of the ZrO2-SiC interface phase, and its change rate is jointly affected by the eddy current effect and the interface polarization effect generated by the electromagnetic field. Preferably, the mapping relationship between the electromagnetic field parameters and the thermal conductivity regulation amplitude is calibrated through experiments, and the following nonlinear model is established: ; wherein, is the magnetic permeability constant of the material, which is related to the guard plate composition and the microstructure of the interface phase; is the reference frequency, which is used to normalize the influence of the electromagnetic field frequency; is the dimensionless exponential coefficient, which reflects the nonlinear action intensity of the electromagnetic parameters on the thermal conductivity regulation.

[0034] The interlayer interface thickness of the gradient guard plate is controlled by a vacuum hot pressing sintering process. Preferably, the deviation of the interface layer thickness uniformity does not exceed a preset threshold to ensure the continuity of heat flow transfer. The introduction of the Cr3C2 / NiCr interface layer can effectively suppress the stress concentration caused by the interlayer thermal expansion mismatch, and at the same time reduce the element interdiffusion between the ZrO2 and SiC phases at high temperatures by forming a diffusion barrier layer.

[0035] The lattice feedback module includes a high-temperature resistant XRD probe that penetrates the guard plate layer and a diffraction spectrum analysis unit. The XRD probe emits X-rays to penetrate the graphitized material and receives the diffraction signal to obtain the crystal plane diffraction data in real time. The diffraction spectrum analysis unit fits the diffraction peak intensity through the Lorentz function, generates a graphitization degree feedback signal and transmits it to the multi-physical field model to correct the material constant in the carbon atom migration dynamics equation; In this embodiment, the lattice feedback module monitors the lattice state of the graphite material in real time through a high-temperature resistant X-ray diffraction (XRD) probe, and feeds back the graphitization degree to the multi-physical field model to achieve parameter self-correction. The specific implementation method is as follows: The high-temperature resistant XRD probe adopts a through-type structure design and includes a tantalum carbide (TaC) protective cover and a liquid metal dynamic sealing layer. The tantalum carbide protective cover covers the front end of the probe and is used to block the deposition of carbon powder in the furnace and thermal radiation damage in a high-temperature environment. Its thickness is optimized according to the X-ray transmittance requirements to meet the preset X-ray transmission threshold. . Preferably, the surface of the protective cover is coated with an aluminum nitride (AlN) thin film to reduce X-ray scattering loss and enhance thermal shock resistance.

[0036] The liquid metal dynamic sealing layer is filled in the internal cavity of the probe and is used to isolate the intrusion of high-temperature gas in the furnace and maintain the cleanliness of the optical path. The liquid metal is composed of a gallium-based alloy, and its composition satisfies , and the absorption coefficient at the X-ray wavelength is , so as to balance the sealing performance and signal attenuation. Preferably, the melting point of the alloy is lower than 50 °C and the boiling point is higher than 2000 °C. The liquid metal is continuously driven to flow by an external circulation pump to maintain the thickness of the sealing layer constant, and the flow rate is adaptively adjusted according to the temperature change in the furnace to avoid sealing failure caused by thermal expansion.

[0037] The working wavelength λ of the XRD probe is selected according to the diffraction characteristics of graphite crystals. Preferably, Cu-Kα rays ( ) are used as the incident source. After being focused by a collimator, they penetrate the graphitized material in the furnace, and the receiving end collects the diffraction signal of the (002) crystal plane. After the diffraction spectrum analysis unit performs background subtraction and noise filtering on the original signal, the Lorentz function is used to fit the intensity of the diffraction peak of the (002) crystal plane , and its expression is: ; Among them, is the peak area; is the full width at half maximum; is the central value of the diffraction angle. Based on the fitting result, the graphitization degree is calculated, and the formula is defined as: ; Among them, is the integral intensity of the full diffraction spectrum ( ), which is calculated in real time through a numerical integration algorithm.

[0038] The graphitization degree feedback signal is transmitted to the multi-physical field model through a data bus and is used to online correct the material kinetic constant in the carbon atom migration dynamics equation. The correction logic is: when the deviation between the measured graphitization degree and the model prediction value exceeds the preset threshold, the parameter optimization algorithm is triggered, and it is iteratively updated based on the least squares method until the deviation converges within the allowable range.

[0039] A multi-objective optimization module, which receives the regulation instruction and the graphitization degree feedback signal, dynamically optimizes the temperature uniformity, crystal orientation degree and unit energy consumption index through a reinforcement learning algorithm, generates a set of process parameters and feeds them back to the dynamic matrix control module to form a closed-loop control; In this embodiment, the multi-objective optimization module dynamically balances the temperature uniformity, crystal orientation degree and energy consumption index through a reinforcement learning algorithm, generates a process parameter optimization instruction and feeds it back to the dynamic matrix control module. The specific implementation method is as follows: The optimization objective function is defined as maximizing the comprehensive benefit index, and its mathematical expression is: ; where, represents the integrated intensity of the diffraction peak of the graphite (002) crystal plane, which is calculated in real time from the diffraction spectrum data output by the lattice feedback module and reflects the crystal orientation degree; is the standard deviation of the axial temperature in the furnace, which is extracted from the fusion data of the temperature monitoring module and characterizes the temperature uniformity; is the unit mass energy consumption, which is calculated according to the ratio of the integrated value of the heating power to the carbon source treatment amount. The calculation formula is: ; where, is the heating power density distribution output by the dynamic matrix control module; is the cumulative mass of the treated carbon source.

[0040] The dynamic weight coefficient satisfies the normalization condition , and its initial value is set according to the process priority. Preferably, a higher weight is given to in the initial stage of graphitization to strengthen the control of the crystal orientation degree, and the weight of is increased in the stable stage to give priority to ensuring the temperature uniformity.

[0041] The weight coefficient is updated online through a reinforcement learning algorithm, and the update rule is: ; where, is the learning rate, and its value range is , which is used to control the weight update step size; is the preset comprehensive benefit target value, which is set according to historical process data and product specifications; is the real-time comprehensive benefit index, which is calculated from the weighted combination of , , , and the expression is: ; Among them, is the theoretical maximum value of the diffraction intensity of the (002) crystal plane, which is used for normalization processing.

[0042] The state space of the reinforcement learning algorithm is defined as , and the action space is the adjustment amount of the weight coefficient , and the reward function is designed as the negative absolute value of the comprehensive benefit deviation: ; Preferably, the policy gradient algorithm is used for policy optimization, and the weight coefficient is iteratively updated by the gradient ascent method to maximize the long-term cumulative reward.

[0043] After the optimized set of process parameters is generated, it is fed back to the dynamic matrix control module through the data bus, triggering the re-optimization of the heating power regulation instruction, and forming a closed-loop control loop. This process can correct the response characteristics of process parameters to multi-objective conflicts in real time, and solve the problem that traditional fixed-weight strategies are difficult to adapt to dynamic process conditions.

[0044] Please refer to Figure 2 , the present invention also provides a method for optimizing the temperature uniformity of a graphitization furnace, and the method includes the following steps: Step S1: Obtain the two-dimensional temperature field distribution data in the furnace through an infrared thermal imager array, and at the same time collect local temperature data through a thermocouple group with an axial gradient distribution, and output multi-source temperature signals; Through an infrared thermal imager array uniformly arranged along the circumferential direction of the furnace body, the two-dimensional temperature field distribution data of the furnace cross-section is synchronously collected in multiple spectral bands, covering the full-region spatial information in the furnace; at the same time, a distributed thermocouple group is arranged in a gradient segmentation along the axial direction of the furnace body and symmetrically divided in the circumferential direction to collect local temperature point data of each temperature zone.

[0045] The infrared thermal imager array and the thermocouple group achieve millisecond-level time synchronization through a high-precision clock signal, eliminate the time sequence deviation of data acquisition, and ensure the spatio-temporal consistency of temperature signals. The measurement end of the thermocouple is embedded under the surface of the furnace insulation layer at a preset depth to avoid the influence of carbon powder pollution on the measurement accuracy.

[0046] Step S2: Weight and fuse the two-dimensional temperature field distribution data collected by the infrared thermal imager and the local temperature data of the thermocouple group according to the temperature zone division to generate equivalent temperature data for each temperature zone; The two-dimensional temperature field data collected by the infrared thermal imager is divided into several sub-regions according to the physical position of the temperature zone, and the spatial average value of the temperature data in each sub-region is taken; at the same time, the local measurement value of the thermocouple in the corresponding temperature zone is extracted, and the two types of data are linearly fused through a preset weight coefficient to generate equivalent temperature data for each temperature zone.

[0047] The weight coefficient is calibrated according to the spatial resolution error of the infrared thermal imager and the local measurement error distribution characteristics of the thermocouple. The infrared data is given a higher weight to retain the global distribution characteristics of the temperature field, while the thermocouple data is used to correct the local measurement deviation.

[0048] Step S3: inputting the equivalent temperature data into a multi-physics field model of the simultaneous temperature field transient heat transfer equation and the carbon atom migration kinetics equation to predict the temperature field distribution and graphitization degree evolution trend in the future time domain; The fused equivalent temperature data is fed into a multiphysics model that combines the transient heat transfer equation for the temperature field with the kinetic equation for carbon atom migration. The temperature field equation describes the dynamic relationship between heating power, carbon source feed rate, and temperature distribution; the kinetic equation correlates the temperature gradient with the rate of graphitization evolution, predicting the temperature field distribution and the material's graphitization progress in the future time domain.

[0049] The model is discretely solved by the finite difference method, and the boundary conditions are iteratively updated in combination with real-time temperature data to improve the prediction accuracy.

[0050] Step S4: Based on the prediction results of the multi-physics field model, the optimal adjustment amount of the heating power of each temperature zone is calculated by a dynamic matrix control algorithm to generate a heating power control instruction; Based on the prediction results of the multi-physics model, a dynamic matrix control algorithm is used to continuously optimize the heating power adjustment for each temperature zone. By constructing a step response matrix to characterize the thermal coupling effect between temperature zones, the objective function is designed to minimize the deviation between the predicted temperature and the target value, while constraining the power adjustment range to a safe range.

[0051] The algorithm solves the optimal control sequence through quadratic programming, generates heating power control instructions and sends them to the actuator.

[0052] Step S5: According to the heating power control instruction, the electron concentration of the interface phase of the ZrO2-SiC gradient guard plate is adjusted by an alternating electromagnetic field, and the axial thermal conductivity distribution of the guard plate is dynamically adjusted to match the temperature gradient direction monitored in real time; Based on control instructions, an alternating electromagnetic field is applied to the ZrO2-SiC gradient shield, altering the electron concentration in the interface phase through eddy currents and dynamically adjusting the shield's axial thermal conductivity. The shield's thermal conductivity is adjusted in the direction of the real-time temperature gradient. When a positive temperature gradient is detected, thermal conductivity is increased to accelerate heat dissipation, while reversed thermal conductivity is reduced to inhibit heat dissipation.

[0053] The electromagnetic field parameters (frequency, intensity) are adaptively adjusted according to the thermal conductivity control requirements to ensure that the thermal properties of the guard plate are adapted to the process conditions in real time.

[0054] Step S6: Obtain the crystal plane diffraction intensity data of the graphite material in real time through a high-temperature-resistant X-ray diffraction probe, analyze and calculate the current graphitization degree, and feed back this data into the carbon atom migration kinetics equation to correct the material kinetics constant online; The high-temperature-resistant X-ray diffraction probe emits X-rays to penetrate the graphitized material in real time, receives the diffraction signal of the (002) crystal plane and analyzes the integrated intensity to calculate the current graphitization degree. The probe adopts a tantalum carbide protective cover and a liquid metal dynamic sealing layer design to maintain the cleanliness of the optical path and the signal stability in a high-temperature environment.

[0055] The graphitization degree data is fed back to the carbon atom migration kinetics equation in real time, and the material kinetics constant is corrected through an online parameter identification algorithm to eliminate the systematic deviation between the model prediction and the measured value.

[0056] Step S7: Synthesize the heating power regulation instruction, the real-time graphitization degree and the energy consumption data, and dynamically balance the temperature uniformity, the crystal orientation degree and the unit energy consumption index through a reinforcement learning algorithm, generate an optimized set of process parameters and feed it back to Step S4 to form a closed-loop control; Synthesize the heating power regulation instruction, the real-time graphitization degree and the energy consumption data, and use a reinforcement learning algorithm to dynamically adjust the weight distribution of the temperature uniformity, the crystal orientation degree and the energy consumption index, and generate an optimized set of process parameters. The algorithm iteratively updates the weight coefficients based on the deviation between the comprehensive benefit target value and the actual value to approach the optimal balance point.

[0057] The optimized parameter set is fed back to Step S4 to trigger the re-optimization of the heating power instruction, forming a "monitoring - prediction - regulation - feedback" closed-loop control loop to realize the adaptive adjustment of the process parameters.

[0058] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made in these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An optimization processing system for the temperature uniformity of a graphitization furnace, characterized in that, The system comprises: The temperature monitoring module includes an infrared thermal imager array and a distributed thermocouple group. The infrared thermal imager array uses multi-spectral acquisition to collect the two-dimensional temperature field distribution in the furnace. The thermocouple group is distributed in the insulation layer of the furnace with axial gradient and circumferential symmetry to output multi-source temperature data. a dynamic matrix control module that receives the multi-source temperature data in real time, inputs the two-dimensional temperature field distribution data into a multi-physics field model that combines the temperature field transient heat transfer equation and the carbon atom migration kinetics equation, and generates heating power control instructions for each temperature zone through calculation using the multi-physics field model; a material-control coupling module comprising a ZrO2-SiC gradient guard plate and an electromagnetic control unit, wherein the electromagnetic control unit regulates the electron concentration of the interface phase of the gradient guard plate by applying an alternating electromagnetic field according to the control instruction, and dynamically adjusts the axial thermal conductivity distribution of the guard plate to match the real-time temperature gradient distribution obtained by the temperature monitoring module; A lattice feedback module includes a high-temperature resistant XRD probe that penetrates the guard plate layer and a diffraction spectrum analysis unit. The XRD probe transmits X-rays through the graphitized material and receives diffraction signals to obtain crystal plane diffraction data in real time. The diffraction spectrum analysis unit fits the diffraction peak intensity using a Lorentz function to generate a graphitization degree feedback signal and transmits it to the multi-physics field model to correct the material constants in the carbon atom migration kinetics equation; The multi-objective optimization module receives the control instructions and the graphitization degree feedback signal, dynamically optimizes the temperature uniformity, crystal orientation and unit energy consumption indicators through a reinforcement learning algorithm, generates a process parameter set and feeds it back to the dynamic matrix control module to form a closed-loop control.

2. The temperature uniformity optimization processing system of a graphitization furnace according to claim 1, characterized in that, In the temperature monitoring module: The infrared thermal imager array collects two-dimensional temperature field distribution data in the furnace in multiple spectral bands , where represents the transverse coordinate; represents the axial coordinate; represents the time variable; The distributed thermocouple group is axially divided into segments and circumferentially divided into zones and evenly arranged. The total number of temperature measurement points is , and the temperature data of each temperature zone is output ; The fusion of multi-source temperature data calculates the equivalent temperature of each temperature zone through the following formula: ； Among them, is the infrared image pixel region corresponding to the temperature zone; is the total number of pixels in the region; and are fusion weight coefficients, satisfying .

3. The temperature uniformity optimization processing system of a graphitization furnace according to claim 1, characterized in that, In the dynamic matrix control module: The multiphysics model consists of the following simultaneous equations: Transient heat transfer equation of temperature field: ； Among them, is the axial temperature distribution in the furnace; is the equivalent thermal diffusion coefficient; is the heating power - temperature conversion coefficient; is the axial heating power density distribution; is the carbon source mass flow rate influence coefficient; is the carbon source feeding rate; Carbon atom migration kinetic equation: ； wherein, is the graphitization degree; is the standard deviation of the axial temperature; is the material kinetic constant.

4. The temperature uniformity optimization processing system of a graphitization furnace according to claim 3, characterized in that, The dynamic matrix control module also includes: Predictive control algorithm: Based on the step response matrix , construct the prediction time domain and the control time domain . The objective function is: ； wherein, is the target temperature distribution; is the temperature tracking weight matrix, and the diagonal element ; is the control increment weight matrix, and the diagonal element ; is the heating power increment vector at the th moment; Solution rule: Solve the optimal control sequence through the quadratic programming algorithm , satisfying: ； Among them and are power increment constraints.

5. The temperature uniformity optimization processing system of a graphitization furnace according to claim 1, characterized in that, The material-control coupling module includes: Thermal conductivity control equation of gradient guard plate: The axial thermal conductivity distribution of the guard plate is dynamically adjusted according to the real-time temperature gradient to meet the following requirements: ； wherein, is the axial thermal conductivity distribution of the guard plate; is the thermal conductivity of the substrate; is the thermal conductivity regulation range; is the sign function, taking +1 when the temperature gradient direction is positive and -1 otherwise; Working rules of electromagnetic control unit: The frequency of the applied alternating electromagnetic field , the magnetic field strength ; The electromagnetic field adjusts the thermal conductivity to meet the requirement by changing the electron concentration at the ZrO2-SiC interface phase , such that: 。 6. The temperature uniformity optimization processing system of a graphitization furnace according to claim 5, characterized in that, The gradient guard plate structure: The ZrO2-SiC gradient protection plate is composed of layers, and the th layer has a ZrO2 volume fraction that satisfies: ； Among them, is the upper limit of the ZrO2 volume fraction at the inlet end of the guard plate; is the lower limit at the outlet end; Adjacent layers are connected by a Cr3C2 / NiCr interface layer, and the interface thickness is controlled within a preset range to reduce interlayer thermal stress and enhance mechanical bonding strength; Electromagnetic control optimization rules: Establish the mapping relationship between the regulation range of thermal conductivity and electromagnetic field parameters: ； wherein, is the material permeability constant; is the reference frequency; is the exponential coefficient.

7. The temperature uniformity optimization processing system of a graphitization furnace according to claim 1, characterized in that, In the lattice feedback module: The high-temperature resistant XRD probe includes a tantalum carbide protective cover and a liquid metal dynamic sealing layer, which is used to emit and receive X-ray signals in a high-temperature environment, where the X-ray wavelength is , and the diffraction spectrum analysis unit fits the intensity of the crystal plane diffraction peak through the Lorentz function , and calculates the graphitization degree according to the following formula : ； Among them, is the integral intensity of the full diffraction spectrum, and the thickness of the tantalum carbide protective cover satisfies the X-ray transmittance , and the thermal expansion coefficient of the liquid metal dynamic sealing layer matches the material of the protective cover.

8. The temperature uniformity optimization processing system of a graphitization furnace according to claim 7, characterized in that, The liquid metal dynamic sealing layer is composed of a gallium-based alloy, and its composition satisfies , at the X-ray wavelength the absorption coefficient , and the thickness of the sealing layer is maintained constant by a circulation pump, and the flow rate is adaptively adjusted according to the temperature in the furnace.

9. The temperature uniformity optimization processing system of a graphitization furnace according to claim 1, characterized in that, In the multi-objective optimization module: The optimization objective function is defined as: ； Among them, is the integrated intensity of the diffraction peak of the crystal plane, calculated from the diffraction peak intensity ratio output by the lattice feedback module; is the standard deviation of the axial temperature in the furnace, obtained in real time through the fusion data of the temperature monitoring module; is the energy consumption per unit mass, calculated based on the integral value of the heating power and the carbon source treatment amount; is the dynamic weight coefficient, satisfying the normalization condition ; The weight coefficient is updated online through the reinforcement learning algorithm, and the update rule is: ； Among them, is the learning rate; is the preset comprehensive benefit target value; is the real-time comprehensive benefit index.

10. A method for optimizing the temperature uniformity of a graphitization furnace, which is applied to the system according to any one of claims 1-9, and is characterized in that, The method comprises the following steps: Step S1: acquiring two-dimensional temperature field distribution data in the furnace through an infrared thermal imager array, and simultaneously collecting local temperature data through an axial gradient-distributed thermocouple group, and outputting a multi-source temperature signal; Step S2: performing weighted fusion on the two-dimensional temperature field distribution data collected by the infrared thermal imager and the local temperature data of the thermocouple group according to the temperature zone division to generate equivalent temperature data of each temperature zone; Step S3: inputting the equivalent temperature data into a multi-physics field model of the simultaneous temperature field transient heat transfer equation and the carbon atom migration kinetics equation to predict the temperature field distribution and graphitization degree evolution trend in the future time domain; Step S4: Based on the prediction results of the multi-physics field model, the optimal adjustment amount of the heating power of each temperature zone is calculated by a dynamic matrix control algorithm to generate a heating power control instruction; Step S5: According to the heating power control instruction, the electron concentration of the interface phase of the ZrO2-SiC gradient guard plate is adjusted by an alternating electromagnetic field, and the axial thermal conductivity distribution of the guard plate is dynamically adjusted to match the temperature gradient direction monitored in real time; Step S6: obtaining the crystal plane diffraction intensity data of the graphite material in real time through a high temperature resistant X-ray diffraction probe, analyzing and calculating the current degree of graphitization, and feeding the data back into the carbon atom migration kinetic equation to correct the material kinetic constants online; Step S7: Combining the heating power control instructions, real-time graphitization degree and energy consumption data, a reinforcement learning algorithm is used to dynamically balance the temperature uniformity, crystal orientation and unit energy consumption indicators, generate an optimized set of process parameters and feed it back to step S4 to form a closed-loop control.

Images