A high precision power control method and system for laser annealing

Abstract

The application relates to the technical field of material processing, and discloses a high-precision power control method and system for laser annealing. The method comprises the following steps: acquiring initial thermal conductivity coefficient data, collecting environmental factor data in real time, and determining a preliminary path model of heat propagation; determining a temperature monitoring threshold according to the preliminary path model; collecting real-time temperature data, calculating a temperature gradient, adjusting a power control parameter after the temperature gradient exceeds the temperature monitoring threshold, optimizing a dynamic representation of the thermal conductivity coefficient, and determining an adaptability coefficient; obtaining a corrected power output sequence according to the adaptability coefficient; calculating a heat conduction rate according to the corrected power output sequence, and calculating an updated path model; determining a final power control scheme according to the updated path model; and optimizing a feedback integration process according to the final power control scheme, and obtaining an optimized sequence for different materials. The method can realize efficient and personalized material processing.

Description

Technical Field

[0001] This invention relates to the field of materials processing technology, and in particular to a high-precision power control method and system for laser annealing. Background Technology

[0002] Currently, laser annealing technology is a key process in materials processing and semiconductor manufacturing. It uses a laser beam to rapidly heat and cool the surface of materials, thereby optimizing and reorganizing the microstructure. It is widely used in the manufacturing process of high-end electronic devices such as smartphone chips and solar cells.

[0003] In a current technology, power control is typically based on preset fixed parameters, which has limited ability to monitor and adjust the temperature gradient and heat conduction rate in real time during annealing. However, since the thermal conductivity of materials is affected by various factors and varies nonlinearly, fixed parameters are difficult to adapt to the dynamic conditions in actual processing, which can easily lead to uneven heat distribution, local overheating, or thermal stress concentration. This makes it difficult to achieve dynamic optimization and adjustment of power parameters, resulting in insufficient processing accuracy and stability.

[0004] In summary, existing technologies suffer from uneven heat distribution in materials during laser annealing. Summary of the Invention

[0005] This invention provides a high-precision power control method and system for laser annealing to solve the problem of uneven heat distribution in materials during laser annealing.

[0006] In a first aspect, to solve the above-mentioned technical problems, the present invention provides a high-precision power control method for laser annealing, comprising: Acquire initial thermal conductivity data and collect environmental factor data in real time to determine a preliminary path model for heat transfer; Based on the preliminary path model, the heat distribution in the material processing area is simulated and calculated to identify potential areas of uneven heat distribution and determine the temperature monitoring threshold range. Real-time temperature data during the laser annealing process is collected and the temperature gradient is calculated. When the temperature gradient exceeds the temperature monitoring threshold range, the power control parameters are adjusted and the adjusted parameters are obtained. The dynamic characterization of thermal conductivity is optimized based on the adjusted power control parameters to determine the adaptability coefficient; Based on the aforementioned adaptability coefficient, the thermal stress distribution data of the simulated material microstructure is used to obtain the corrected power output sequence; Based on the corrected power output sequence, the heat conduction rate is iteratively calculated, and when the fluctuation amplitude of the heat conduction rate exceeds a preset fluctuation threshold, the environmental factor data is re-collected and the preliminary path model is updated to obtain the updated path model. Based on the updated path model, real-time temperature monitoring data is integrated to generate a closed-loop control cycle, and the final power control scheme is determined. The heat distribution uniformity index is extracted from the final power control scheme, and the feedback integration process is optimized based on the heat distribution uniformity index to obtain the optimized sequence for different materials.

[0007] Secondly, the present invention provides a high-precision power control system for laser annealing, comprising: The data acquisition module is used to obtain initial thermal conductivity data and collect environmental factor data in real time to determine the preliminary path model of heat transfer. The threshold determination module is used to simulate and calculate the heat distribution in the material processing area based on the preliminary path model, identify potential areas of uneven heat distribution, and determine the temperature monitoring threshold range. The power control module is used to collect real-time temperature data during the laser annealing process and calculate the temperature gradient. When the temperature gradient exceeds the temperature monitoring threshold range, the power control parameters are adjusted and the adjusted parameters are obtained. The optimization and control module is used to optimize the dynamic characterization of thermal conductivity and determine the adaptability coefficient based on the adjusted power control parameters. The simulation calculation module is used to simulate the thermal stress distribution data of the material microstructure based on the fitness coefficient to obtain the corrected power output sequence; The heat conduction calculation module is used to iteratively calculate the heat conduction rate based on the corrected power output sequence, and when the fluctuation amplitude of the heat conduction rate exceeds a preset fluctuation threshold, re-collect the environmental factor data and update the preliminary path model to obtain the updated path model. The iterative update module is used to integrate real-time temperature monitoring data to generate a closed-loop control cycle based on the updated path model, and to determine the final power control scheme. The closed-loop control module is used to extract the heat distribution uniformity index from the final power control scheme, and optimize the feedback integration process based on the heat distribution uniformity index to obtain the optimized sequence for different materials.

[0008] Thirdly, the present invention also provides an electronic device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement a high-precision power control method for laser annealing as described in any one of the above.

[0009] Fourthly, the present invention also provides a computer-readable storage medium comprising a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to perform a high-precision power control method for laser annealing as described in any one of the above.

[0010] Compared with the prior art, the present invention has the following beneficial effects: (1) By acquiring the initial thermal conductivity data in material processing and collecting environmental factor data in real time, and combining finite element analysis and finite difference method to establish a heat propagation path model, the present invention can accurately predict the uneven heat distribution area, thereby realizing dynamic monitoring and threshold control of temperature gradient in laser annealing process, and solving the problem of uneven heat distribution of materials in laser annealing process in the prior art.

[0011] (2) This invention introduces a genetic algorithm to dynamically characterize and optimize the thermal conductivity, and combines a particle swarm optimization algorithm to refine the feedback integration process, thereby realizing personalized control of power control parameters, improving the system's adaptability to different material processing conditions, and enhancing the stability and processing accuracy of the laser annealing process.

[0012] (3) By constructing a closed-loop control cycle, integrating real-time temperature monitoring data and environmental parameters, the present invention iteratively corrects the power output sequence and updates the path model to form an adaptive power control scheme, thereby optimizing the thermal stress distribution at the microstructure level and finally generating an optimized sequence for different materials, which improves the uniformity and reliability of the laser annealing process. Attached Figure Description

[0013] Figure 1 This is a schematic flowchart of a high-precision power control method for laser annealing provided in the first embodiment of the present invention; Figure 2 This is a schematic diagram of a high-precision power control system for laser annealing provided in the second embodiment of the present invention. Detailed Implementation

[0014] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0015] Reference Figure 1 The first embodiment of the present invention provides a high-precision power control method for laser annealing, comprising the following steps: S11: Obtain initial thermal conductivity data and collect environmental factor data in real time to determine the preliminary path model for heat transfer; S12, Based on the preliminary path model, simulate and calculate the heat distribution in the material processing area, identify potential areas with uneven heat distribution, and determine the temperature monitoring threshold range; S13, collect real-time temperature data during the laser annealing process and calculate the temperature gradient. When the temperature gradient exceeds the temperature monitoring threshold range, adjust and obtain the adjusted power control parameters. S14, optimize the dynamic characterization of thermal conductivity based on the adjusted power control parameters, and determine the adaptability coefficient; S15, Based on the adaptability coefficient, simulate the thermal stress distribution data of the material microstructure to obtain the corrected power output sequence; S16, Based on the corrected power output sequence, iteratively calculate the heat conduction rate, and when the fluctuation amplitude of the heat conduction rate exceeds a preset fluctuation threshold, re-collect the environmental factor data and update the preliminary path model to obtain the updated path model; S17. Based on the updated path model, integrate real-time temperature monitoring data to generate a closed-loop control cycle and determine the final power control scheme. S18, extract the heat distribution uniformity index from the final power control scheme, and optimize the feedback integration process based on the heat distribution uniformity index to obtain the optimization sequence for different materials.

[0016] In step S11, acquiring initial thermal conductivity data and collecting environmental factor data in real time to determine a preliminary path model for heat propagation includes: S111, Obtain initial thermal conductivity data during material processing; S112, environmental factor data is collected in real time by sensors, and mean filtering and noise reduction processing is performed on the environmental factor data to obtain humidity change data. When the humidity change data exceeds a preset humidity change threshold, the initial thermal conductivity data is dynamically corrected to obtain the corrected thermal conductivity. S113. Based on the corrected thermal conductivity, the heat propagation path is calculated using the finite difference method to obtain the preliminary path model.

[0017] In step S111, the initial thermal conductivity data during material processing is obtained.

[0018] Initial thermal conductivity data can be obtained using a heat flow meter method. The sample is placed between two plates, and a constant unidirectional heat flow is introduced perpendicularly to the plates. A calibrated sensor is then placed between the plates and the sample to measure the heat flow through the sample. Once the temperatures of the hot and cold plates have stabilized, the sample thickness, the temperatures of the upper and lower surfaces, and the heat flow through the sample are measured. The thermal conductivity of the sample is then calculated using Fourier's law. When the processed material is a thin film (thickness < 1 μm), a scanning thermal microscope or the 3ω method should be used to measure the micro-area thermal conductivity.

[0019] In step S112, environmental factor data is collected in real time by sensors, and the environmental factor data is subjected to mean filtering and noise reduction processing to obtain humidity change data. When the humidity change data exceeds a preset humidity change threshold, the initial thermal conductivity data is dynamically corrected to obtain the corrected thermal conductivity.

[0020] Real-time acquisition of environmental factor data can be achieved using a high-precision humidity sensor. The sensor is installed in the material processing environment to monitor the relative humidity in the air in real time, obtaining the environmental factor data. The environmental factor data is then subjected to mean filtering and noise reduction processing to obtain humidity change data. When the humidity change data exceeds a preset humidity change threshold, the initial thermal conductivity data is dynamically corrected to adjust the material's thermal conductivity, resulting in a corrected thermal conductivity.

[0021] The mean filtering method uses a simple moving average, which involves selecting an odd-sized window and moving it one position forward from the beginning of the sequence. For each position covered by the window, the average of all data within that window is calculated; this average value represents the filtered sequence at the center of the window. At the beginning and end of the sequence, the window cannot completely cover the data; therefore, only values ​​within the complete window are output.

[0022] It should be noted that the choice of window size should be determined based on the sensor sampling frequency and the rate of environmental change. For example, when sampling once per second, a window size of 3 is more appropriate.

[0023] For example, the humidity change data sequence is [65.2%, 65.8%, 70.1%, 64.9%, 65.3%]. By applying a mean filter with a window size of 3, the smoothed data sequence is calculated to be [65.5%, 65.7%, 65.3%]. This method effectively reduces the impact of sensor noise, improves data stability, and provides a more reliable input for subsequent thermal conductivity correction.

[0024] The preset humidity change threshold can be set to 10%. The temperature threshold is set based on experimental statistics. A humidity change of ±10% is the upper limit of common industrial environmental fluctuations, which is sufficient to cause a change of >1% in the thermal conductivity of non-metallic materials.

[0025] When dynamically correcting the initial thermal conductivity data, the thermal conductivity should be adjusted based on the experimentally measured relationship between humidity and thermal conductivity. For example, when humidity increases, the thermal conductivity of the material will decrease due to the adsorption of moisture on the material surface, thus reducing the thermal conductivity.

[0026] For example, when the humidity suddenly increases from 50% to 65%, exceeding the threshold, experiments show that a 15% change in humidity reduces the material's thermal conductivity by 0.8 W / m·K. Therefore, the thermal conductivity is adjusted from 50 W / m·K to 49.2 W / m·K. Here, the humidity change is the absolute value of the difference between the current humidity and the humidity of the previous sampling period. In practice, the correction needs to consider the material's hygroscopic isotherm characteristics, employing piecewise linear or exponential correction. The humidity-thermal conductivity relationship varies greatly among different materials (silicon, ceramics, polymers), and the correction coefficients for different materials need to be calibrated through pre-experiments to establish a fitting formula for Δk=f(ΔH, material_type). Where Δk represents the change in thermal conductivity (unit: W / m·K), indicating the adjustment value of thermal conductivity due to humidity changes, and can be positive or negative; ΔH represents the change in humidity (unit: %RH), defined as the absolute value of the difference between the current humidity and the humidity of the previous sampling period, representing the instantaneous fluctuation range of ambient humidity; material_type represents the material type identifier, used to distinguish different materials (such as silicon, ceramics, polymers) or material groups, because the effect of humidity on thermal conductivity varies with the microstructure, hygroscopicity, and other characteristics of the material. In practice, material_type can be represented as an integer code (such as 0=silicon, 1=ceramics) or a string label, used to index the calibrated parameter set.

[0027] In step S113, the heat propagation path is calculated using the finite difference method based on the corrected thermal conductivity to obtain the preliminary path model.

[0028] Based on the corrected thermal conductivity and the temperature data acquired by the sensor, the material is divided into a mesh using the finite difference method. The heat transfer direction and rate at each mesh point are calculated, and the preliminary path model is obtained using the heat transfer direction and rate at each mesh point. It should be noted that when dividing the material into a mesh using the finite difference method, a non-uniform mesh is used. In regions with drastic temperature changes, near heat sources, or at boundaries, the mesh is denser; in the interior of the material where temperature changes are gradual, the mesh is sparser.

[0029] It is worth noting that explicit time progression and central difference space discretization are employed, and the discretization equation is as follows: ; in, This represents the temperature of the i-th grid node at the nth time step; This represents the temperature of the i-th grid node at the (n+1)-th time step. This represents the thermal conductivity of the material (corrected from step S112). Indicates the time step; Indicates the density of the material; This indicates the specific heat capacity of a material at constant pressure. Indicates the size of the spatial grid.

[0030] Set the laser irradiation surface as a third type of boundary condition: ; in, This represents the temperature gradient along the boundary normal direction; Indicates laser heat flux density; Indicates the convective heat transfer coefficient; Indicates the boundary surface temperature; Indicates ambient temperature.

[0031] When the temperature gradient exceeds At that time, the grid size is reduced to 1 / 5 of the original size.

[0032] In step S12, based on the preliminary path model, the heat distribution in the material processing area is simulated and calculated to identify potential areas of uneven heat distribution and determine the temperature monitoring threshold range. This includes: S121, The preliminary path model is meshed using the finite element analysis method to obtain mesh temperature data; S122, calculate the temperature gradient in the grid temperature data. When the temperature gradient exceeds a preset temperature gradient threshold, adjust the boundary conditions of the grid temperature data to obtain heat distribution data. S123, the heat distribution data is subjected to mean filtering and noise reduction processing to obtain the denoised heat distribution data; S124, The k-means clustering algorithm is used to analyze the denoised thermal distribution data to determine the temperature monitoring threshold range.

[0033] In step S121, the preliminary path model is meshed using the finite element analysis method to obtain mesh temperature data.

[0034] For example, the processing material is aluminum alloy, the processing area is a 100mm × 100mm plane, and the mesh adopts a uniform distribution structure, dividing the mesh into 10,000 small cells, each cell being 1mm × 1mm in size. After meshing, the temperature value of each cell is calculated using finite element analysis software to generate mesh temperature data.

[0035] It should be noted that the accuracy of the mesh generation directly affects the accuracy of the analysis results. An overly coarse mesh may overlook local high-temperature points, while an overly fine mesh will increase the computational cost.

[0036] Preferably, a 1mm grid size strikes a balance between accuracy and efficiency, making it suitable for capturing subtle changes in heat distribution during processing.

[0037] In step S122, the temperature gradient in the grid temperature data is calculated. When the temperature gradient exceeds a preset temperature gradient threshold, the boundary conditions of the grid temperature data are adjusted to obtain heat distribution data.

[0038] The temperature gradient is the temperature difference between a point and its adjacent points, where the adjacent points are defined as two adjacent grids after grid division.

[0039] It should be noted that the specific value of the preset temperature gradient threshold needs to be determined based on the material's thermal stress limit, which is the stress corresponding to the maximum temperature difference the material can withstand. Exceeding this limit will lead to cracking or deformation. For example, the preset temperature gradient threshold can be set to 5-20°C / mm. When processing aluminum alloy, the thermal stress limit of aluminum alloy is approximately 100-200MPa, corresponding to a temperature gradient threshold of 10-15°C / mm.

[0040] The iterative calculation process includes: scanning the temperature gradient to identify regions exceeding the temperature gradient threshold; adjusting the boundary conditions of these regions to reduce the temperature gradient; and repeating the above process until no region exceeds the temperature gradient threshold. The convergence criterion is that the maximum temperature change between two consecutive iterations is less than 0.1°C or the maximum number of iterations (1000) is reached.

[0041] It should be noted that the boundary conditions are adjusted based on the corresponding convective heat transfer boundary conditions in the finite element model, and their mathematical expression is as follows: ; This describes the heat conservation at a tiny point on the material surface. It can be understood as: the heat conducted from the material's interior to its surface is equal to the heat convected from the surface to the cooling fluid. Wherein, Thermal conductivity, For temperature gradient, Surface temperature, This refers to the coolant temperature. is the convective heat transfer coefficient.

[0042] For example, assuming a grid point has a temperature of 120°C and an adjacent point has a temperature of 100°C, the calculated temperature gradient is 20°C / mm, exceeding the threshold. In this case, by iteratively adjusting the boundary conditions, such as increasing the local cooling rate to simulate the coolant jet effect, the temperature gradient is reduced to 8°C / mm. This adjustment is based on the heat conduction model of finite element analysis, ensuring that the heat distribution tends to be uniform.

[0043] Among these adjustments, the boundary conditions are improved by increasing the convective heat transfer coefficient in the formula. Or lower the coolant temperature To achieve this.

[0044] In step S123, the heat distribution data is subjected to mean filtering and denoising processing to obtain denoised heat distribution data.

[0045] It should be noted that the window size should be determined based on the sensor sampling frequency and the rate of environmental change. For example, when sampling once per second, a window size of 3 is more suitable. The mean filtering process described in this step is the same as in step S112, so it will not be elaborated further.

[0046] For example, assuming the acquired temperature data sequence is [102.3°C, 103.1°C, 101.8°C, 104.2°C, 102.5°C], after applying a mean filter with a window size of 3, the denoised sequence is [102.4°C, 103.0°C, 102.8°C]. This method reduces the impact of sensor noise and improves the stability of temperature data through averaging.

[0047] In step S124, the k-means clustering algorithm is used to analyze the denoised thermal distribution data to determine the temperature monitoring threshold range.

[0048] When using the k-means clustering algorithm to classify thermal distribution data, the processing area can be divided into high-temperature zone, medium-temperature zone, and low-temperature zone.

[0049] It should be noted that the initial centroids for k-means clustering are selected using the K-means++ algorithm, which includes: Randomly select the first cluster center from the data points; for each point in the dataset, calculate its distance D(x) to the nearest selected cluster center; according to the probability of D(x)², that is, the farther the point is, the greater the probability of it being selected as the next center, randomly select the next cluster center in a weighted manner; repeat the second and third steps until K initial centers are selected.

[0050] It should be noted that the choice of initial centroids in k-means clustering affects the classification results. Ideally, random initialization and multiple iterations should be used to ensure stable results. This classification method provides data support for subsequent process optimization and helps to accurately identify areas with abnormal heat distribution.

[0051] After clustering is completed, it is necessary to quantitatively evaluate the results to verify whether the choice of K value is reasonable and whether the clustering is effective. The elbow rule can be used to calculate the intra-cluster sum of squares for different K values ​​(e.g., K=1 to K=5). The intra-cluster sum of squares is the sum of the squared distances from each data point to the center of its cluster, decreasing as K increases. Plot the relationship between K value and intra-cluster sum of squares. When the curve shows a clear "inflection point," the K value corresponding to that point is usually the optimal number of clusters. If K=3 is at the inflection point, it proves that dividing into three clusters is reasonable.

[0052] The temperature monitoring threshold range is set according to the temperature of the processing center.

[0053] For example, assuming the temperature distribution data ranges from 80°C to 120°C, and k=3, a clustering algorithm divides the temperature into three intervals: 80-90°C, 90-105°C, and 105-120°C. After classification, high-temperature areas may be concentrated in the processing center, while low-temperature areas are located at the edges, thus determining the temperature monitoring threshold. If the high-temperature area requires focused monitoring, the temperature monitoring threshold is set to 110°C. 110°C is the center point of the high-temperature area after k-means clustering.

[0054] In step S13, real-time temperature data during the laser annealing process is collected and the temperature gradient is calculated. When the temperature gradient exceeds the temperature monitoring threshold range, the adjusted power control parameters are obtained, including: S131, acquire the real-time temperature data collected by the sensor during the laser annealing process, and perform mean filtering and noise reduction processing on the real-time temperature data to obtain the noise-reduced temperature data. S132, calculate the temperature gradient based on the denoised temperature data, and when the temperature gradient exceeds the temperature monitoring threshold range, record the over-threshold region to obtain the over-threshold region data. S133, Based on the data of the super-threshold region, the laser annealing region is divided into partitions to obtain the adjusted heat distribution data; S134, Based on the adjusted heat distribution data, adjust the power parameters of the laser annealing equipment to obtain the adjusted power control parameters.

[0055] In step S131, real-time temperature data during the laser annealing process collected by the sensor is acquired, and the real-time temperature data is subjected to mean filtering and noise reduction processing to obtain the denoised temperature data.

[0056] Real-time acquisition of temperature and humidity data by sensors is crucial for ensuring uniform heat distribution. The mean filtering window size is set to 3. Subsequent mean filtering processing is consistent with step S112, and therefore will not be elaborated further. It should be noted that the window size selection needs to be adjusted according to the sampling frequency and noise level. For example, when sampling once per second, a window size of 3 can balance the noise reduction effect with data real-time performance.

[0057] For example, a silicon wafer with a processing area of ​​50mm × 50mm is used. The sensor collects data once per second, and the temperature data sequence is [150.2°C, 151.5°C, 149.8°C, 152.3°C]. This data is affected by environmental noise and may contain random fluctuations, thus requiring noise reduction. For the above temperature sequence, a window size of 3 is selected, and the denoised sequence is calculated to be [150.5°C, 151.2°C, 151.0°C]. This method is simple and efficient, effectively smoothing the fluctuating data collected by the sensor.

[0058] In step S132, the temperature gradient is calculated based on the denoised temperature data. When the temperature gradient exceeds the temperature monitoring threshold range, the over-threshold region is recorded to obtain the over-threshold region data.

[0059] The temperature gradient reflects the rate of temperature change between adjacent grid points, and is usually expressed in °C / mm.

[0060] For example, assuming the grid is divided into 1mm × 1mm cells, and the temperature at a certain grid point is 151°C, while the temperature at an adjacent point is 148°C, according to the definition of gradient, the temperature gradient is calculated as 3°C / mm using 151°C - 148°C. If the temperature monitoring threshold range is 2°C / mm, then this area is recorded as an over-threshold region.

[0061] The data for the region exceeding the threshold typically includes grid coordinates and corresponding gradient values, which facilitates subsequent analysis and adjustment.

[0062] In step S133, the laser annealing region is divided into partitions based on the data of the region exceeding the threshold, and the adjusted heat distribution data is obtained.

[0063] Based on the data of the over-threshold region, the laser annealing region is divided into sub-regions using a mesh generation method, refining the processing area into multiple sub-regions. Adjusted thermal distribution data is then generated using finite element analysis software. Uniform mesh generation is used in this process.

[0064] For example, a 50mm × 50mm region was divided into 2500 1mm × 1mm grids, with the over-threshold region concentrated at the laser irradiation center. By adjusting boundary conditions, such as increasing the cooling airflow rate to improve the convective heat transfer coefficient, the temperature of the high-temperature region could be reduced. The adjusted heat distribution data was generated using finite element analysis software, showing that the temperature gradient decreased from 3°C / mm to 1.5°C / mm.

[0065] The boundary conditions are adjusted based on the corresponding convective heat transfer boundary conditions in the finite element model, and their mathematical expression is as follows: ; This describes the heat conservation at a tiny point on the material surface. It can be understood as: the heat conducted from the material's interior to its surface is equal to the heat convected from the surface to the cooling fluid. Wherein, Thermal conductivity, For temperature gradient, Surface temperature, This refers to the coolant temperature. is the convective heat transfer coefficient.

[0066] In step S134, the power parameters of the laser annealing equipment are adjusted according to the adjusted heat distribution data to obtain the adjusted power control parameters.

[0067] In one possible implementation, a PID control algorithm can be used to adjust the power parameters of the annealing equipment.

[0068] Specific implementation methods include: Calculate the deviation between the current temperature gradient and the target temperature gradient in real time: ; Where G_current is the currently measured temperature gradient value, and G_target is the target temperature gradient value; Calculate the power adjustment based on the PID control law: ; Where K_p, Ki, and K_d are the proportional, integral, and derivative coefficients, respectively; the PID parameters can be tuned through step response experiments, for example, using K_p=0.5, Ki=0.1, and K_d=0.05 as initial values. The feedforward compensation function can be approximated by a linear model, such as... , where α is the material correlation coefficient, calibrated through a humidity-power experiment. This represents the baseline humidity value, which is the set standard ambient humidity reference value.

[0069] For example, to compensate for the increased thermal resistance, the feedforward compensation is calculated based on real-time collected ambient humidity data using a pre-established humidity-power relationship model. ; in, This is the current humidity value. This is the compensation function calibrated through experiments.

[0070] Set the adjusted power parameters as follows: ; in The adjusted power control parameters, Indicates the amount of power adjustment. These are the initial power control parameters.

[0071] For example, assuming the initial laser power is 500W, after detecting an area exceeding the threshold, if the temperature gradient in a certain area increases to 3.5°C / mm, the PID control algorithm calculates the power adjustment amount. Meanwhile, the humidity sensor detected that the ambient humidity increased from 45% to 50%, and the feedforward compensation model calculated that power compensation was needed. The PID algorithm then adjusts the power to 460W and increases the cooling airflow to make the heat distribution more even.

[0072] In step S14, the step of optimizing the dynamic characterization of thermal conductivity based on the adjusted power control parameters and determining the adaptability coefficient includes: S141, Obtain a feedback signal from the adjusted power control parameters, filter the feedback signal to obtain an integrated signal; S142, Based on the integrated signal, the thermal conductivity is iteratively calculated using a genetic algorithm. The iteration stops when the calculation result converges or reaches the preset number of iterations, thus obtaining dynamic characterization data. S143, when the dynamic characterization data does not match the preset thermal conductivity threshold range, the iteration parameters of the genetic algorithm are adjusted and the calculation is re-performed to obtain the optimized thermal conductivity. S144, the adaptability coefficient is calculated based on the optimized thermal conductivity.

[0073] In step S141, a feedback signal is obtained from the adjusted power control parameters, and the feedback signal is filtered to obtain an integrated signal.

[0074] The feedback signal is obtained from the adjusted power control parameters, and decomposed into different frequency components using wavelet transform. Specifically, this involves: selecting the Daubechies 4 (db4) wavelet as the basis function, which is suitable for processing non-stationary temperature signals; performing a 4-level wavelet decomposition, where the number of decomposition levels is determined based on the signal sampling frequency, using the decomposition level L = log2(fs / 2f_c), where fs is the sampling frequency and f_c is the signal cutoff frequency; and performing soft thresholding denoising, using a general thresholding strategy. , where σ is the noise standard deviation and N is the signal length; the denoised wavelet coefficients are reconstructed to obtain the filtered integrated signal.

[0075] The feedback signals include real-time data on laser power, temperature distribution, and environmental parameters, with a sampling frequency of no less than 100Hz to ensure the capture of dynamic temperature changes. These signals may be affected by equipment vibration (frequency range 5-200Hz) or environmental noise (frequency range 0.1-10Hz).

[0076] In step S142, based on the integrated signal, the thermal conductivity is iteratively calculated using a genetic algorithm. The iteration stops when the calculation result converges or reaches the preset number of iterations, thus obtaining dynamic characterization data. The specific parameter settings for the genetic algorithm are as follows: The population size is set to 50-100 individuals based on the problem complexity, with a recommended population size of 80 for silicon wafers; the initial mutation rate is set to 0.1, dynamically adjusted based on convergence; the crossover probability is set to 0.8, using arithmetic crossover; roulette wheel selection is used, with a selection pressure coefficient of 2; and the fitness function is set to... Where wit1=0.7 and wit2=0.3 are weighting coefficients, RMSE is the root mean square error between simulated and observed temperatures, and |Δk| is a penalty term for the change in thermal conductivity. When the rate of change of the optimal fitness is less than 0.1% for 10 consecutive generations, it is considered relatively convergent; when the fitness value reaches 0.95 or higher, it is considered absolutely convergent; convergence stops when the maximum number of iterations (100 generations) is reached.

[0077] For example, assuming the initial thermal conductivity is set to 50 W / m·K, the genetic algorithm, through crossover and mutation operations with a mutation rate of 0.1, obtains dynamic characterization data of 52.3 W / m·K after 10 iterations.

[0078] In step S143, when the dynamic characterization data does not match the preset thermal conductivity threshold range, the iteration parameters of the genetic algorithm are adjusted and the calculation is performed again to obtain the optimized thermal conductivity.

[0079] For example, if the dynamic characterization data is 52.3 W / m·K and the preset thermal conductivity threshold range is 48-50 W / m·K, the mutation rate can be adjusted, such as setting 0.1 to 0.15, and the genetic algorithm in step S142 can be iterated again. The parameters of the genetic algorithm, such as the population size and mutation rate, need to be adjusted according to the material type and annealing process.

[0080] For example, experimental verification shows that setting the population size of the silicon wafer to 100 balances computational efficiency and accuracy. The mutation rate can be set to 0.1. If the dynamic characterization data does not match the preset thermal conductivity threshold range, the iteration parameters of the genetic algorithm are adjusted to increase the mutation rate. When adjusting the iteration parameters, overfitting must be avoided to ensure the stability of the calculation results.

[0081] It is worth noting that if the dynamic characterization data does not match the preset thermal conductivity threshold range (e.g., 48-50 W / m·K), the parameters should be adjusted according to the following strategy: If the convergence speed is too slow: increase the mutation rate from 0.1 to 0.15-0.2; if premature convergence occurs: increase the population size by 20% and reinitialize; if the oscillation is severe: reduce the crossover probability to 0.6-0.7.

[0082] S144, the adaptability coefficient is calculated based on the optimized thermal conductivity.

[0083] The calculation requires comprehensive consideration of thermal conductivity, temperature gradient, and ambient humidity to generate an adaptability coefficient to guide equipment parameter adjustments. The specific calculation steps include: normalizing thermal conductivity, temperature gradient, and ambient humidity; and merging the three normalized parameters into a comprehensive state index representing the overall system state, according to the formula: ; in , , These are weighting coefficients, determined through principal component analysis. For example, historical data is collected to calculate the covariance matrix, and the first principal component loadings are extracted as weights. A typical value is... =0.5, =0.3, =0.2, , , , which are the normalized thermal conductivity, temperature gradient, and ambient humidity, respectively. The comprehensive state index is mapped to an adaptability coefficient directly used to control the equipment, according to the formula: ; Here, λ is the sensitivity coefficient, which controls how quickly α changes with S_index. We can set λ = 1.

[0084] For example, with a thermal conductivity of 49.8 W / m·K, a current humidity of 48%, and a temperature gradient of 2°C / mm, the fitness coefficient can be calculated to be 0.95.

[0085] In step S15, based on the adaptability coefficient, the thermal stress distribution data of the simulated material microstructure is used to obtain the corrected power output sequence, including: S151, Obtain initial thermal stress distribution data from a pre-established device physical model; S152, Based on the adaptability coefficient, the initial thermal stress distribution data is simulated and calculated to obtain the simulated thermal stress distribution; S153, extract feedback data from the simulated thermal stress distribution, and perform wavelet transform on the feedback data to obtain a filtered signal; S154, when the filtered signal does not match the preset performance threshold, the particle swarm optimization algorithm is used to iteratively optimize the simulation parameters to obtain thermal stress distribution data; S155, Based on the thermal stress distribution data, the corrected power output sequence is calculated.

[0086] In step S151, initial thermal stress distribution data is obtained from the pre-established device physical model.

[0087] The physical model of a device typically includes the physical properties of the material, such as the coefficient of thermal expansion, Young's modulus, and Poisson's ratio. These physical properties can be obtained through experimental measurement. For example, the coefficient of thermal expansion can be measured using laser interferometry, which utilizes the principle of laser interferometry to measure the length change of the material during heating; Young's modulus can be measured using a tensile method, by stretching the material to measure the relevant parameters; and Poisson's ratio can be measured by applying an axial tensile force to induce axial strain in the sample.

[0088] The initial thermal stress distribution data can be obtained from the heat conduction equation and the mechanical equilibrium equation.

[0089] For example, the coefficient of thermal expansion of a silicon wafer is The Young's modulus is 130 GPa, and the initial thermal stress distribution data indicates that the maximum thermal stress in the processing area is 150 MPa.

[0090] In step S152, based on the adaptability coefficient, the initial thermal stress distribution data is simulated and calculated using the finite element analysis method to obtain the simulated thermal stress distribution.

[0091] The geometric model of the device is imported into the finite element analysis software. The material properties obtained from the physical model of the device are assigned to the geometric model. The geometric model is discretized into a large number of small elements. Initial thermal stress distribution data is input, and the simulated thermal stress distribution is calculated by the finite element analysis software.

[0092] For example, 1000 elements are divided on a 50mm × 50mm silicon wafer using eight-node hexahedral elements (SOLID185), with a central region of 0.5mm and an edge region of 2mm, and a transition gradient of 1.2. After assigning values ​​to temperature-related material parameters, boundary conditions are set, including thermal boundary conditions, mechanical boundary conditions, and environmental convection. The Newton-Raphson iterative method is used, with the convergence criterion set as the residual. The maximum number of iterations was 50, and the time step was set to 0.001s (explicit dynamic analysis). The local stress of each element was calculated based on the initial thermal stress data. The simulation results showed that the maximum simulated thermal stress distribution in the machining center area reached 160MPa, while that in the edge area was 120MPa.

[0093] In step S153, feedback data is extracted from the simulated thermal stress distribution, and wavelet transform is performed on the feedback data to separate noise and effective signals, thereby obtaining the filtered signal.

[0094] The feedback data extracted from the simulated thermal stress distribution includes the material's stress values, temperature gradient, and displacement data. The wavelet basis function was set to Daubechies 8 (db8); the number of decomposition layers was set to 5, corresponding to a frequency range of 0-31.25Hz; the Stein unbiased risk estimation (SURE) threshold was used; subsequently, the layer thresholds were set to 0.4 for high-frequency layers and 0.2 for low-frequency layers.

[0095] For example, the stress values ​​of the material in the feedback data are [160MPa, 158MPa, 162MPa]. Wavelet transform processes the feedback data into a smooth filtered signal [159.5MPa, 159.8MPa, 159.6MPa], which preserves the stress change trend and reduces the impact of noise.

[0096] In step S154, when the filtered signal does not match the preset performance threshold, the particle swarm optimization algorithm is used to iteratively optimize the simulation parameters to obtain thermal stress distribution data.

[0097] The preset performance thresholds include the stress value of the material. This design index is derived by comprehensively weighing the material's function, safety, cost, and reliability. The stress value of the material is required to be between 140-150 MPa.

[0098] For example, a particle swarm optimization algorithm is used, with the population size set to 30 particles; the learning factor is set to... The inertia weight is set to linearly decreasing, with a maximum value of 0.9 and a minimum value of 0.4; the maximum number of iterations is set to 50; the mutation probability is set to 0.1, using Gaussian mutation. Relative convergence is defined as the optimal fitness change rate being <0.1% for 10 generations; absolute convergence is defined as the fitness value being <0.01; iteration stops when the maximum number of generations or computation time limit is reached. During optimization, the stress threshold is set to 140-150 MPa (based on material yield strength), the temperature gradient threshold is ≤100 K / mm, and the safety factor is 1.5-2.0.

[0099] The initial thermal conductivity coefficient was set to 50 W / m·K, and after 10 iterations, it was optimized to 48.5 W / m·K. Thermal stress distribution data was generated, and the stress in the central region was reduced to 145 MPa, which met the threshold requirements.

[0100] In S155, the corrected power output sequence is calculated based on the thermal stress distribution data.

[0101] Based on the thermal stress distribution data, the device is divided into a finite number of units, and the stress response of each unit is calculated to obtain the corrected power output sequence. During the calculation, a PID control algorithm can be used, with input parameters including stress distribution and environmental parameters such as humidity. The PID parameters are tuned to K_p=2.5, Ki=0.1, K_d=0.05; integral separation is set as follows: when... The integration stops at MPa; the output is limited to ±20% of the rated power; the sampling period is set to 0.1s, and a power adjustment scheme is generated.

[0102] For example, assuming the initial laser power is 450W and the humidity is 45%, the PID control algorithm calculates a power correction coefficient of 0.98, and then the laser power is adjusted from 450W to 441W. This adjustment ensures a more uniform stress distribution.

[0103] In step S16, the process of iteratively calculating the heat conduction rate based on the corrected power output sequence, and re-collecting the environmental factor data and updating the preliminary path model when the fluctuation amplitude of the heat conduction rate exceeds a preset fluctuation threshold, to obtain the updated path model, includes: S161, obtain initial data of heat conduction rate from the modified power output sequence, and perform iterative calculation on the initial data using the finite difference method. Stop the iteration when the calculation result converges or reaches the preset number of calculations to obtain the heat conduction rate sequence. S162, The heat conduction rate sequence is processed and transformed to obtain the fluctuation amplitude; S163, if the fluctuation amplitude exceeds the preset fluctuation threshold, the environmental factor data is re-collected to obtain an environmental dataset; S164. Based on the environmental dataset, the particle swarm optimization algorithm is used to optimize the preliminary path model to obtain the updated path model.

[0104] In step S161, initial data of the heat conduction rate is obtained from the modified power output sequence, and the initial data is iteratively calculated using the finite difference method. The iteration stops when the calculation result converges or the preset number of calculations is reached, thus obtaining the heat conduction rate sequence.

[0105] The initial data for obtaining the thermal conductivity rate needs to be based on the physical device model and environmental processing parameters. For example, in a silicon-based laser processing scenario, the initial thermal conductivity rate data can be generated from the corrected power output sequence by combining the thermal conductivity equation with material properties. The thermal conductivity of a silicon wafer is typically 148 W / m·K, and combined with the temperature gradient of the processing area, such as 50 K / mm, the initial thermal conductivity rate can be calculated to be 7400 W / m².

[0106] When iteratively calculating the initial data using the finite difference method, the device is divided into a discrete grid, and the heat flux change at each grid point is calculated. The heat flux change at each grid point constitutes the heat conduction rate sequence.

[0107] In step S162, the heat conduction rate sequence is processed and transformed to obtain the fluctuation amplitude.

[0108] The rate fluctuation data, including peak values, trough values, and frequency of change, are extracted from the heat conduction rate sequence. Statistical analysis methods are used to calculate the standard deviation and mean of the rate fluctuation data to obtain the fluctuation amplitude. The frequency of change can be determined using zero-crossing rate detection, calculating the number of times the signal crosses the mean; that is, the frequency of change equals the number of zero-crossings divided by the time window length.

[0109] For example, the heat conduction rate sequence of the processing center area is [7400W / m², 7200W / m², 7600W / m²]. Calculating its range, the maximum variation of the fluctuation data is 400W / m², and the fluctuation range is 350W / m².

[0110] In step S163, if the fluctuation amplitude exceeds the preset fluctuation threshold, the environmental factor data is re-collected to obtain an environmental dataset.

[0111] If the fluctuation amplitude exceeds a preset fluctuation threshold, the environmental factor data, including humidity, ambient temperature, and airflow speed in the processing area, will be re-collected using sensors in the device. This data will be recorded to obtain an environmental dataset. The environmental dataset provides an environmental basis for subsequent optimization and helps identify the causes of excessive fluctuations.

[0112] The preset fluctuation threshold is based on collecting at least 30 sets of fluctuation data under normal process conditions, covering different materials and different power parameters. According to statistical methods, the fluctuation threshold can be set as the average fluctuation amplitude plus the standard deviation of three fluctuation amplitudes.

[0113] In step S164, based on the environmental dataset, the particle swarm optimization algorithm is used to optimize the preliminary path model to obtain the updated path model.

[0114] It should be noted that the path model describes the impact of the laser processing path on the heat conduction rate, and the parameters in the particle swarm optimization algorithm include the path velocity and the laser focus point. The population size is set at 20-50 particles. The learning factor is set to c11=c21=1.496; the inertia weight is set to 0.4-0.9, decreasing linearly; the optimization objective is set to minimize the temperature gradient variance, and the constraint condition is set to thermal stress < allowable value, which refers to the maximum thermal stress threshold that the material can withstand, exceeding this value may lead to material failure; the number of iterations is set to 50-100 times; the fitness function is set in the same way as the particle swarm optimization algorithm mentioned above, and will not be elaborated further; the convergence condition is fitness change <0.1% for 10 generations or reaching the maximum number of iterations.

[0115] In one alternative implementation, the initial path speed can be set to 10 mm / s, and the focal point diameter to 0.1 mm. The number of iterations can be set to 20, which can achieve the desired optimization effect without consuming too many resources.

[0116] In step S17, based on the updated path model, real-time temperature monitoring data is integrated to generate a closed-loop control cycle, and the final power control scheme is determined, including: S171, receives real-time temperature monitoring data from temperature monitoring equipment; S172, Based on the updated path model, the real-time temperature monitoring data is processed to obtain an integrated dataset; S173, Based on the integrated dataset, the model parameters of the updated path model are dynamically adjusted to obtain a set of control parameters; S174, when the parameter values ​​of the control parameter set exceed the preset control threshold, the operating status of the equipment is analyzed to obtain a status dataset; S175, Based on the state dataset, optimize the power control scheme and determine the final power control scheme.

[0117] In step S171, real-time temperature monitoring data is received from the temperature monitoring device.

[0118] The system receives real-time temperature monitoring data from temperature monitoring devices, which are typically deployed at key locations within the processing area. For example, in laser processing of silicon-based materials, the equipment can collect real-time temperature data of the processed surface using an infrared thermometer.

[0119] In step S172, the real-time temperature monitoring data is processed according to the updated path model to obtain an integrated dataset.

[0120] Based on the updated path model, the real-time temperature data and environmental parameters are normalized using a weighted average or data fusion method. Then, a comprehensive thermal impact factor is calculated according to preset data fusion weights. This comprehensive thermal impact factor is combined with spatial location information to obtain an integrated dataset. The environmental parameters include ambient temperature, humidity, and airflow velocity.

[0121] It should be noted that the preset data fusion weights need to be obtained through regression analysis based on historical data. In step S173, the model parameters of the updated path model are dynamically adjusted according to the integrated dataset to obtain a set of control parameters. The comprehensive thermal impact factor and its spatial location information are obtained from the integrated dataset. Using a closed-loop control algorithm, when the algorithm detects the presence of a thermal impact factor, a real-time feedback mechanism dynamically adjusts the path model parameters, including path velocity and laser power, to obtain the set of control parameters.

[0122] In one alternative implementation, the closed-loop control algorithm can use a PID control algorithm, which is optimized in combination with the speed of the processing path and the laser power.

[0123] Specifically, the first step is to analyze and evaluate the comprehensive thermal impact factors in the integrated dataset. When the monitoring data deviates from the preset process target, the system's closed-loop control algorithm activates its response mechanism. This algorithm not only responds proportionally to the current deviation but also accumulates historical deviations through integral action to eliminate steady-state errors and uses differential elements to predict trends to suppress system overshoot. Through multi-dimensional analysis and synthesis of the deviation, the algorithm outputs a comprehensive parameter correction. This correction is precisely allocated to executable processing parameters, such as the coordinated adjustment of laser power and scanning speed, via preset mapping rules within the system. For example, when the system determines that the thermal impact factor in the processing center area exceeds the ideal threshold, it generates a set of control parameters that include reducing laser power and adjusting motion speed, thereby achieving precise control of the processing heat input.

[0124] For example, with an initial path speed of 10 mm / s and a laser power of 50 W, after closed-loop control, the path speed is adjusted to 9.5 mm / s and the power is adjusted to 48 W. These parameters constitute the aforementioned control parameter set, which can effectively adapt to thermal changes during processing and improve processing stability.

[0125] In step S174, when the parameter values ​​of the control parameter set exceed the preset control threshold, the operating status of the equipment is analyzed to obtain a status dataset.

[0126] The state detection method involves monitoring operating parameters of the equipment, such as vibration frequency and current fluctuations, using sensors. The operating state of the equipment is then recorded to obtain the state dataset.

[0127] For example, the sensor shows that the device vibration frequency is 50Hz and the current fluctuation amplitude is 0.2A, indicating that the device is in a slightly overloaded state. The status dataset records this data, providing comprehensive information about the device's operation for subsequent optimization.

[0128] In step S175, the power control scheme is optimized based on the state dataset to determine the final power control scheme.

[0129] Among them, when optimizing the power control scheme, a genetic algorithm can be used to simulate the evolution process of the parameter space and gradually optimize the laser power and pulse frequency.

[0130] Specifically, the genetic algorithm uses key parameters of the power control scheme, such as laser power and pulse frequency, as decision variables, and establishes a multi-dimensional optimization model under multiple constraints, including stable operating range of the equipment, processing quality requirements, and energy efficiency. By simulating the iterative computation process of natural selection or swarm intelligence, the algorithm gradually filters out the Pareto optimal solution from a vast parameter space that simultaneously satisfies the following requirements: controlling the equipment load within a safe threshold while ensuring processing accuracy, and maintaining a dynamic balance of heat input. After multiple rounds of iteration and convergence, the system generates a recalibrated final power control scheme. For example, using the genetic algorithm, the initial laser power is set to 50W, the initial pulse frequency to 100kHz, the maximum number of iterations to 10, the population size to 100, the roulette wheel selection method, and the mutation rate to 0.1. After 10 iterations, the optimized scheme is a laser power of 48W and a pulse frequency of 95kHz. This scheme effectively balances processing efficiency and equipment stability, ensuring the reliability of the processing.

[0131] The initial power and initial pulse frequency are determined based on the performance parameters of the annealing equipment.

[0132] In step S18, the extraction of heat distribution uniformity index from the final power control scheme and the optimization of the feedback integration process based on the heat distribution uniformity index to obtain an optimization sequence for different materials include: S181, Obtain temperature distribution data from the final power control scheme, calculate the root mean square error of the temperature distribution data, and obtain the heat distribution uniformity index; S182, Based on the heat distribution uniformity index, the particle swarm optimization algorithm is used to iteratively calculate the residual deviation. When the calculation result converges or reaches the preset number of optimizations, the iteration stops, and the optimization parameter set is obtained. S183, when the execution parameter value in the optimization parameter set exceeds the preset optimization threshold, analyze the relationship between the material properties and the temperature distribution data to obtain the state parameter set; S184, extract the feedback results of the state parameter set, analyze and integrate the feedback results, and generate optimization sequences for different materials.

[0133] In step S181, temperature distribution data is obtained from the final power control scheme, the root mean square error of the temperature distribution data is calculated, and the heat distribution uniformity index is obtained.

[0134] The mean square error method quantifies the uniformity of heat distribution by calculating the sum of squares of the deviations between the temperature at each measuring point and the average temperature, thus obtaining a heat distribution uniformity index. This index directly reflects the degree of heat concentration during the processing.

[0135] In step S182, based on the heat distribution uniformity index, the particle swarm optimization algorithm is used to iteratively calculate the residual deviation. The iteration stops when the calculation result converges or reaches the preset number of optimizations, and the optimized parameter set is obtained.

[0136] Based on the aforementioned heat distribution uniformity index, a particle swarm optimization algorithm is employed to simulate group cooperative behavior, search for the optimal solution in the parameter space, and iteratively calculate the residual deviation. Iteration stops when the calculation results converge or the preset number of optimization iterations is reached, resulting in the optimized parameter set. It should be noted that the number of iterations and the population size of the particle swarm optimization algorithm need to be reasonably set according to processing requirements to balance computational efficiency and optimization accuracy.

[0137] In silicon wafers, the population size can be set to 100 and the number of iterations can be set to 10. This parameter can balance computational efficiency and optimization accuracy.

[0138] For example, the initial laser power was 52W and the processing speed was 9.8mm / s. After multiple iterations, it was optimized to a laser power of 50W and a processing speed of 9.6mm / s. These parameters constitute an optimized parameter set, which helps to balance heat input and processing efficiency.

[0139] In step S183, when the execution parameter values ​​in the optimization parameter set exceed the preset optimization threshold, the relationship between the material properties and the temperature distribution data is analyzed, the execution parameter values ​​are adjusted, and a state parameter set is obtained.

[0140] When the execution parameter values ​​in the optimized parameter set exceed a preset optimization threshold, the changes in thermal stress on the material surface and the material properties after processing are monitored to obtain a state parameter set. This state parameter set provides comprehensive information on the material behavior for subsequent adjustments. Changes in thermal stress on the material surface are detected using a spectrometer, or the material properties after processing are evaluated using a hardness tester.

[0141] For example, if the detection shows that the thermal stress in the processed area is 120 MPa and the material hardness has slightly decreased to 90% of the initial value, it indicates that high temperature may cause changes in material properties. The state parameter set will record the state information at this time.

[0142] In step S184, the feedback results of the state parameter set are extracted, and the feedback results are analyzed and integrated to generate an optimization sequence for different materials.

[0143] Specifically, by using parameters from a weighted state parameter set, including fusion temperature, material properties, and environmental parameters, optimized sequences for different materials are generated.

[0144] In one optional implementation, the equipment extracts multi-dimensional feedback data such as temperature, material properties, and environmental parameters. Using a weighted data fusion method, it calculates a comprehensive index that reflects the current processing status. Based on this comprehensive index, macro-control schemes such as "heat reduction," "energy increase," and "shape stabilization" are formed. Using a process knowledge base, the macro-control schemes are precisely mapped to a set of specific equipment parameters that can work together, such as laser power and pulse frequency, resulting in the final optimized sequence.

[0145] For example, for silicon-based materials, the optimized sequence suggests reducing the laser power to 49W and adjusting the pulse frequency to 90kHz to accommodate the material properties at high temperatures. For silicon carbide materials, the optimized sequence might suggest a power of 51W and a frequency of 92kHz.

[0146] In summary, this invention discloses a high-precision power control method for laser annealing, which solves the problem of uneven heat distribution during material processing.

[0147] Reference Figure 2 The second embodiment of the present invention provides a high-precision power control system for laser annealing, comprising: The data acquisition module is used to obtain initial thermal conductivity data and collect environmental factor data in real time to determine the preliminary path model of heat transfer. The threshold determination module is used to simulate and calculate the heat distribution in the material processing area based on the preliminary path model, identify potential areas of uneven heat distribution, and determine the temperature monitoring threshold range. The power control module is used to collect real-time temperature data during the laser annealing process and calculate the temperature gradient. When the temperature gradient exceeds the temperature monitoring threshold range, the power control parameters are adjusted and the adjusted parameters are obtained. The optimization and control module is used to optimize the dynamic characterization of thermal conductivity and determine the adaptability coefficient based on the adjusted power control parameters. The simulation calculation module is used to simulate the thermal stress distribution data of the material microstructure based on the fitness coefficient to obtain the corrected power output sequence; The heat conduction calculation module is used to iteratively calculate the heat conduction rate based on the corrected power output sequence, and when the fluctuation amplitude of the heat conduction rate exceeds a preset fluctuation threshold, re-collect the environmental factor data and update the preliminary path model to obtain the updated path model. The iterative update module is used to integrate real-time temperature monitoring data to generate a closed-loop control cycle based on the updated path model, and to determine the final power control scheme. The closed-loop control module is used to extract the heat distribution uniformity index from the final power control scheme, and optimize the feedback integration process based on the heat distribution uniformity index to obtain the optimized sequence for different materials.

[0148] It should be noted that the high-precision power control device for laser annealing provided in this embodiment of the invention is used to execute all the process steps of the high-precision power control method for laser annealing in the above embodiment. The working principles and beneficial effects of the two are one-to-one, so they will not be described again.

[0149] This invention also provides an electronic device. The electronic device includes a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a high-precision power control program for laser annealing. When the processor executes the computer program, it implements the steps in the various embodiments of the high-precision power control method for laser annealing described above, for example... Figure 1 The step S11 shown. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the above-described device embodiments, such as the optimization and control module.

[0150] For example, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the electronic device.

[0151] The electronic device may be a desktop computer, laptop, handheld computer, or smart tablet, etc. The electronic device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the above components are merely examples of electronic devices and do not constitute a limitation on the electronic device. It may include more or fewer components than described above, or combine certain components, or different components. For example, the electronic device may also include input / output devices, network access devices, buses, etc.

[0152] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the electronic device, connecting all parts of the electronic device via various interfaces and lines.

[0153] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0154] Wherein, if the modules / units integrated in the electronic device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0155] It should be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0156] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A high-precision power control method for laser annealing, characterized in that, include: Acquire initial thermal conductivity data and collect environmental factor data in real time to determine a preliminary path model for heat transfer; Based on the preliminary path model, the heat distribution in the material processing area is simulated and calculated to identify potential areas of uneven heat distribution and determine the temperature monitoring threshold range. Real-time temperature data during the laser annealing process is collected and the temperature gradient is calculated. When the temperature gradient exceeds the temperature monitoring threshold range, the power control parameters are adjusted and the adjusted parameters are obtained. The dynamic characterization of thermal conductivity is optimized based on the adjusted power control parameters to determine the adaptability coefficient; Based on the aforementioned adaptability coefficient, the thermal stress distribution data of the simulated material microstructure is used to obtain the corrected power output sequence; Based on the corrected power output sequence, the heat conduction rate is iteratively calculated, and when the fluctuation amplitude of the heat conduction rate exceeds a preset fluctuation threshold, the environmental factor data is re-collected and the preliminary path model is updated to obtain the updated path model. Based on the updated path model, real-time temperature monitoring data is integrated to generate a closed-loop control cycle, and the final power control scheme is determined. The heat distribution uniformity index is extracted from the final power control scheme, and the feedback integration process is optimized based on the heat distribution uniformity index to obtain the optimized sequence for different materials.

2. The high-precision power control method for laser annealing according to claim 1, characterized in that, The process of acquiring initial thermal conductivity data and collecting environmental factor data in real time to determine a preliminary path model for heat propagation includes: Obtain initial thermal conductivity data during material processing; Environmental factor data is collected in real time by sensors, and mean filtering and noise reduction are performed on the environmental factor data to obtain humidity change data. When the humidity change data exceeds the preset humidity change threshold, the initial thermal conductivity data is dynamically corrected to obtain the corrected thermal conductivity. Based on the corrected thermal conductivity, the heat propagation path is calculated using the finite difference method to obtain the preliminary path model.

3. The high-precision power control method for laser annealing according to claim 1, characterized in that, The step of simulating and calculating the heat distribution in the material processing area based on the preliminary path model, identifying potential areas of uneven heat distribution, and determining the temperature monitoring threshold range includes: The preliminary path model was meshed using the finite element method to obtain mesh temperature data. Calculate the temperature gradient in the grid temperature data. When the temperature gradient exceeds a preset temperature gradient threshold, adjust the boundary conditions of the grid temperature data to obtain heat distribution data. The heat distribution data is subjected to mean filtering and denoising to obtain denoised heat distribution data; The denoised thermal distribution data is analyzed using the k-means algorithm to determine the temperature monitoring threshold range.

4. The high-precision power control method for laser annealing according to claim 1, characterized in that, The process involves acquiring real-time temperature data during laser annealing and calculating the temperature gradient. When the temperature gradient exceeds the temperature monitoring threshold range, the power control parameters are adjusted to obtain the adjusted parameters, including: The real-time temperature data collected by the sensor during the laser annealing process is acquired, and the real-time temperature data is subjected to mean filtering and noise reduction processing to obtain the noise-reduced temperature data. The temperature gradient is calculated based on the denoised temperature data. When the temperature gradient exceeds the temperature monitoring threshold range, the over-threshold region is recorded to obtain the over-threshold region data. Based on the data of the over-threshold region, the laser annealing region is divided into partitions to obtain adjusted thermal distribution data; Based on the adjusted heat distribution data, the power parameters of the laser annealing equipment are adjusted to obtain the adjusted power control parameters.

5. The high-precision power control method for laser annealing according to claim 1, characterized in that, The step of optimizing the dynamic characterization of thermal conductivity and determining the adaptability coefficient based on the adjusted power control parameters includes: The feedback signal is obtained from the adjusted power control parameters, and the feedback signal is filtered to obtain the integrated signal. Based on the integrated signal, a genetic algorithm is used to iteratively calculate the thermal conductivity. The iteration stops when the calculation result converges or reaches the preset number of iterations, thus obtaining dynamic characterization data. When the dynamic characterization data does not match the preset thermal conductivity threshold range, the iteration parameters of the genetic algorithm are adjusted and the calculation is re-performed to obtain the optimized thermal conductivity. The adaptability coefficient is calculated based on the optimized thermal conductivity.

6. The high-precision power control method for laser annealing according to claim 1, characterized in that, The step of simulating the thermal stress distribution data of the material's microstructure based on the fitness coefficient to obtain the corrected power output sequence includes: Initial thermal stress distribution data are obtained from a pre-established device physical model; Based on the adaptability coefficient, the initial thermal stress distribution data are simulated and calculated to obtain the simulated thermal stress distribution; Feedback data is extracted from the simulated thermal stress distribution, and wavelet transform is performed on the feedback data to obtain a filtered signal; When the filtered signal does not match the preset performance threshold, the particle swarm optimization algorithm is used to iteratively optimize the simulation parameters to obtain thermal stress distribution data. Based on the thermal stress distribution data, the corrected power output sequence is calculated.

7. The high-precision power control method for laser annealing according to claim 1, characterized in that, The step involves iteratively calculating the heat conduction rate based on the corrected power output sequence, and when the fluctuation amplitude of the heat conduction rate exceeds a preset fluctuation threshold, re-collecting the environmental factor data and updating the initial path model to obtain an updated path model, including: Initial data of heat conduction rate are obtained from the modified power output sequence. The initial data is iteratively calculated using the finite difference method. The iteration stops when the calculation result converges or the preset number of calculations is reached, thus obtaining the heat conduction rate sequence. The heat conduction rate sequence is processed and transformed to obtain the fluctuation amplitude; If the fluctuation amplitude exceeds the preset fluctuation threshold, the environmental factor data will be collected again to obtain an environmental dataset. Based on the aforementioned environmental dataset, the initial path model is optimized using the particle swarm optimization algorithm to obtain the updated path model.

8. The high-precision power control method for laser annealing according to claim 1, characterized in that, The step of integrating real-time temperature monitoring data to generate a closed-loop control cycle based on the updated path model, and determining the final power control scheme, includes: Receive real-time temperature monitoring data from temperature monitoring equipment; Based on the updated path model, the real-time temperature monitoring data is processed to obtain an integrated dataset; Based on the integrated dataset, the model parameters of the updated path model are dynamically adjusted to obtain a set of control parameters; When the parameter values ​​of the control parameter set exceed the preset control threshold, the operating status of the equipment is analyzed to obtain a status dataset. Based on the state dataset, the power control scheme is optimized to determine the final power control scheme.

9. The high-precision power control method for laser annealing according to claim 1, characterized in that, The step of extracting a heat distribution uniformity index from the final power control scheme and optimizing the feedback integration process based on the heat distribution uniformity index to obtain an optimization sequence for different materials includes: Temperature distribution data is obtained from the final power control scheme, and the root mean square error of the temperature distribution data is calculated to obtain the heat distribution uniformity index. Based on the heat distribution uniformity index, the particle swarm optimization algorithm is used to iteratively calculate the residual deviation. The iteration stops when the calculation result converges or reaches the preset number of optimizations, and the optimization parameter set is obtained. When the execution parameter values ​​in the optimization parameter set exceed the preset optimization threshold, the relationship between the material properties and the temperature distribution data is analyzed to obtain the state parameter set; The feedback results of the state parameter set are extracted, analyzed and integrated, and an optimization sequence for different materials is generated.

10. A high-precision power control system for laser annealing, characterized in that, include: The data acquisition module is used to obtain initial thermal conductivity data and collect environmental factor data in real time to determine the preliminary path model of heat transfer. The threshold determination module is used to simulate and calculate the heat distribution in the material processing area based on the preliminary path model, identify potential areas of uneven heat distribution, and determine the temperature monitoring threshold range. The power control module is used to collect real-time temperature data during the laser annealing process and calculate the temperature gradient. When the temperature gradient exceeds the temperature monitoring threshold range, the power control parameters are adjusted and the adjusted parameters are obtained. The optimization and control module is used to optimize the dynamic characterization of thermal conductivity and determine the adaptability coefficient based on the adjusted power control parameters. The simulation calculation module is used to simulate the thermal stress distribution data of the material microstructure based on the fitness coefficient to obtain the corrected power output sequence; The heat conduction calculation module is used to iteratively calculate the heat conduction rate based on the corrected power output sequence, and when the fluctuation amplitude of the heat conduction rate exceeds a preset fluctuation threshold, re-collect the environmental factor data and update the preliminary path model to obtain the updated path model. The iterative update module is used to integrate real-time temperature monitoring data to generate a closed-loop control cycle based on the updated path model, and to determine the final power control scheme. The closed-loop control module is used to extract the heat distribution uniformity index from the final power control scheme, and optimize the feedback integration process based on the heat distribution uniformity index to obtain the optimized sequence for different materials.

Images