Brittle material laser cutting thermal stress control method and system

Abstract

The invention discloses a brittle material laser cutting thermal stress control method and system, and the method comprises the steps: obtaining real-time temperature distribution through infrared thermal imaging, combining finite element analysis and material characteristic dynamic data, constructing a thermal stress and fracture relation model, carrying out the iterative calculation of a thermal stress out-of-control threshold through a lambda algorithm, and determining a regulation boundary condition. According to the boundary condition, laser parameters are optimized, synchronous cooling airflow is introduced, the temperature gradient is dynamically controlled through an airflow pressure adjusting algorithm, historical data are processed through a convolutional neural network, the optimal laser parameter combination is predicted, in the cutting process, the microcrack diffusion state is monitored in real time, and auxiliary technological parameters are finely adjusted when necessary. The problem of fracture caused by out-of-control thermal stress in laser cutting of the brittle material is effectively solved, the cutting quality and efficiency are improved, and a new technical scheme is provided for precision machining of the brittle material.

Description

Technical Field

[0001] The present invention relates to the field of intelligent processing technology, and in particular to a method and system for controlling thermal stress in laser cutting of brittle materials. Background Art

[0002] Fully automatic laser cutting technology for brittle materials occupies a vital position in modern manufacturing and is widely used in fields such as semiconductors, glass products and ceramic processing. Its high precision and high efficiency play an irreplaceable role in promoting the development of industrial intelligence.

[0003] With the increasing requirements for product quality and processing accuracy, how to effectively control the characteristics of brittle materials during the cutting process has become one of the key research directions in this field. Traditional processing methods often find it difficult to strike a balance between efficiency and material integrity, resulting in frequent brittle fractures or microcracks, which seriously restricts processing quality and subsequent applications. Existing solutions mostly rely on single process adjustments, such as simply increasing laser power or changing cutting speed, but these methods often have obvious limitations. Simply increasing power may cause excessive heat input, leading to local stress concentration and microcrack propagation in the material; relying solely on speed optimization is difficult to adapt to the needs of different material properties, and the effect is unstable. The fundamental flaw of these methods is the lack of systematic regulation of the thermal stress distribution and fracture mechanism of brittle materials, making it difficult to solve the brittleness problem from the root.

[0004] The core challenges of fully automated laser cutting lie in three key technical factors: managing thermal stress, controlling temperature gradients, and precisely matching laser parameters. Failure to effectively control thermal stress can lead to microcracks in the material due to heat accumulation during the cutting process. Excessively steep temperature gradients can induce thermal shock, exacerbating brittleness. Furthermore, the dynamic relationship between laser parameters and material properties remains largely unresolved, making it difficult to predict optimal cutting conditions. These unresolved technical issues directly raise the challenge of maintaining material toughness and preventing fracture during high-speed cutting, creating a bottleneck in improving processing quality.

[0005] Therefore, how to optimize the laser action mode and auxiliary process means, accurately control thermal stress and temperature gradient, and establish a predictive model of brittle material properties and laser parameters to reduce the risk of brittle fracture during the cutting process has become a key issue that needs to be urgently addressed in this study. Summary of the Invention

[0006] In order to solve the problems raised in the above background technology, the first aspect of the present invention provides a method for controlling thermal stress in laser cutting of brittle materials, the method comprising: S1, obtains real-time temperature distribution data of brittle materials during laser cutting. The surface temperature changes of the material are captured by infrared thermal imaging technology. Based on the influence of heat accumulation and steep temperature gradient, the temperature gradient value is calculated to obtain the initial thermal distribution characteristics. S2, extract the local thermal stress concentration area from the initial thermal distribution characteristics, use the finite element analysis method to simulate the thermal stress distribution under the thermal stress runaway state, combine the dynamic data of material properties, determine the potential location and range of microcrack propagation, and determine the thermal stress concentration coefficient; S3, based on the thermal stress concentration coefficient and temperature gradient value, construct a mathematical model of the relationship between thermal stress and material fracture, introduce dynamic parameters of material properties, and iteratively calculate the thermal stress runaway threshold through the lambda algorithm to obtain the thermal stress control boundary conditions; S4, adjust the laser parameter matching scheme according to the thermal stress control boundary conditions, obtain the laser power and pulse frequency combination that dynamically adapts to the material properties through the pre-established laser energy distribution database, and determine the optimized laser parameters; S5, through the optimized laser parameters and combined with auxiliary process optimization technology, introduces synchronous cooling airflow during the cutting process to control the steepness of the temperature gradient, and uses the airflow pressure regulation algorithm to dynamically adjust the airflow intensity to obtain a gentle temperature gradient distribution; S6, extracting the probability of thermal shock induction from the gentle temperature gradient distribution. If the probability of thermal shock induction exceeds a preset threshold value of thermal shock induction probability, the cutting speed adjustment parameter is adjusted to reduce the influence of heat accumulation, determine the degree of reduction in real-time fracture risk, and determine stable cutting conditions; S7, based on the stable cutting conditions and dynamic data of material properties, trains the parameter prediction model, uses a convolutional neural network to process the historical data of temperature distribution, thermal stress distribution and laser parameters, and outputs the optimal laser parameter combination that matches the characteristics of the brittle material; S8, after obtaining the optimal laser parameter combination, is applied in real time in the fully automatic laser cutting system. The microcrack diffusion state is monitored by sensors. If the microcrack diffusion area exceeds the preset microcrack diffusion area threshold, the auxiliary process optimization parameters are fine-tuned to obtain the final cutting quality control solution.

[0007] Optionally, step S1, obtaining real-time temperature distribution data of the brittle material during laser cutting, capturing the surface temperature change of the material using infrared thermal imaging technology, calculating the temperature gradient value based on the influence of heat accumulation and the steep temperature gradient, and obtaining the initial thermal distribution characteristics, includes: Step S11, collecting surface temperature data of the brittle material during laser cutting by infrared thermal imaging technology to obtain real-time temperature distribution; Step S12, using a two-dimensional convolution algorithm to process the real-time temperature distribution and extract the temperature change trend; Step S13, calculating the temperature gradient of the heat accumulation area based on the extracted temperature change trend to obtain a gradient value distribution; Step S14: using the principal component analysis method to obtain the characteristic vector of the initial thermal distribution based on the gradient value distribution, and determining the influence range of the thermal accumulation; Step S15, if the temperature gradient in the feature vector exceeds the temperature gradient threshold, the initial thermal distribution is smoothed by mean filtering to obtain an optimized thermal distribution; Step S16, extracting distribution features from the optimized thermal distribution to determine the thermal stress concentration area; Step S17: classify the thermal stress concentration areas using a support vector machine algorithm, use radial basis function as a kernel function, train a classification model, and determine the potential crack location.

[0008] Optionally, step S17, classifying the thermal stress concentration area by a support vector machine algorithm, using a radial basis function as a kernel function, training a classification model, and determining the potential crack location, includes: Step S171 , collecting temperature distribution of thermal stress concentration areas through thermal imaging data to obtain preliminary area division results; Step S172: grouping the preliminary region division results using a K-means clustering algorithm to determine the boundary range of the thermal stress concentration; Step S173, extracting spatial feature data from the boundary range to obtain a feature vector of thermal stress distribution; Step S174, classifying the feature vectors using a support vector machine algorithm, using a radial basis function as a kernel function, and training a classification model; Step S175: If the confidence level output by the classification model exceeds the confidence threshold, a high-risk area for potential cracks is determined; Step S176: For high-risk areas, a sliding window method is used to obtain the local temperature change trend and determine the crack propagation direction; Step S177: extract dynamic features from the crack propagation direction and determine the priority order of crack development.

[0009] Optionally, step S3 constructs a mathematical model of the relationship between thermal stress and material fracture based on the thermal stress concentration coefficient and the temperature gradient value, introduces dynamic parameters of material properties, and iteratively calculates the thermal stress runaway threshold through the lambda algorithm to obtain the thermal stress control boundary conditions, including: Step S31, based on the thermal stress concentration and temperature gradient data, a mathematical model is constructed using the finite element method to calculate the thermal stress value distribution; Step S32: inputting the thermal stress value distribution result into the Laplace iterative algorithm for iterative calculation to determine the dynamic parameter change trend; Step S33, adjusting the thermal stress value using the material constitutive relationship according to the dynamic parameter change trend; Step S34: if the adjusted thermal stress value exceeds the preset thermal stress runaway threshold, recalculate the control boundary using a mathematical model; Step S35, by regulating the boundary to limit the thermal stress value, using the fracture mechanics criterion to determine the conditions for material fracture; Step S36, obtaining the concentrated distribution of thermal stress under the material fracture condition, and iteratively updating the thermal stress out-of-control threshold using the least squares method; Step S37: Using the updated thermal stress runaway threshold, determine the final control boundary condition.

[0010] Optionally, step S4, adjusting the laser parameter matching scheme based on the thermal stress control boundary conditions, obtaining a laser power and pulse frequency combination dynamically adapted to the material properties through a pre-established laser energy distribution database, and determining the optimized laser parameters, includes: Step S41, obtaining initial values ​​of laser parameters related to boundary conditions through a pre-established laser energy distribution database to obtain a preliminary matching solution; Step S42: In the preliminary matching scheme, material characteristic data is introduced, dynamic adaptation characteristics are extracted, and the corresponding laser power range is determined; Step S43, after determining the laser power range, using a binary search algorithm to adjust the pulse frequency to obtain a frequency combination that matches the laser power; Step S44: If the thermal stress control requirements change, re-match the laser parameters and update the matching solution based on the energy distribution data in the database; Step S45, obtaining laser parameters that match the boundary conditions according to the updated matching scheme, and determining an adjusted thermal stress control scheme; Step S46: In the thermal stress control scheme, linear regression is used to analyze the correlation between material properties and laser power to determine the optimization direction of production parameters; Step S47: According to the parameter optimization direction, the ratio of pulse frequency to laser power is adjusted to determine the final laser parameters.

[0011] Optionally, in step S46, in the thermal stress control scheme, linear regression is used to analyze the correlation between material properties and laser power to determine the direction of production parameter optimization, including: Step S461: Acquire characteristic data related to thermal stress regulation through a pre-established material property database to determine an initial analysis range; Step S462: For the initial analysis range, linear regression is used to analyze the correlation between material characteristics and laser power to obtain characteristic change trends; Step S463: extracting a production parameter set that matches the control scheme from the energy distribution data based on the characteristic change trend, and determining the production parameter screening direction; Step S464: If the production parameter screening direction deviates from the expectation, the material characteristic data is updated through the database to obtain an adjusted characteristic set; Step S465: cluster analysis is performed on the adjusted feature set to prioritize production parameter adjustments and determine an optimized production parameter combination. Step S466, adjusting the boundary conditions of the control scheme by optimizing the production parameter combination to obtain a balanced scheme for thermal stress distribution; Step S467: According to the balancing solution, the final production parameter ratio is extracted from the matching solution to determine the execution parameters of the thermal stress control.

[0012] Optionally, step S6 extracts the thermal shock induction probability from the gentle temperature gradient distribution. If the thermal shock induction probability exceeds a preset thermal shock induction probability threshold, the cutting speed adjustment parameter is adjusted to reduce the influence of heat accumulation, and the degree of reduction in real-time fracture risk is determined to determine the stable cutting conditions, including: Step S61, obtaining temperature gradient distribution data through a sensor, and calculating the probability of thermal shock induction using kernel density estimation to obtain a probability value; Step S62: If the probability value exceeds the thermal shock induction probability threshold, the relationship between thermal shock and heat accumulation is predicted by linear regression to determine the cutting speed adjustment range; Step S63, updating the cutting speed parameter according to the adjustment range, using a real-time monitoring tool to obtain the trend of heat accumulation changes, and determining the degree of heat accumulation reduction; Step S64, calculating the fracture risk change rate based on the relationship between the degree of heat accumulation reduction and the fracture risk, to obtain a risk reduction index; Step S65: If the risk reduction index reaches the stable threshold, the stable cutting state is confirmed by the ARIMA model to determine the validity of the current cutting speed parameter; Step S66, obtaining real-time fracture risk data, comparing the risk reduction index with the stability threshold, and determining whether further adjustment of the cutting speed parameters is required; Step S67, by iterating the above process in a loop, the temperature gradient data is updated using kernel density estimation to obtain the optimized stable cutting conditions.

[0013] Optionally, step S7, based on the stable cutting conditions and dynamic data of material properties, trains a parameter prediction model, uses a convolutional neural network to process historical data of temperature distribution, thermal stress distribution, and laser parameters, and outputs an optimal laser parameter combination that matches the brittle material properties, including: Step S71, processing historical data through a convolutional neural network, inputting historical data as time series temperature and thermal stress data, using the TensorFlow framework for training, and obtaining initial laser parameter prediction results; Step S72, extracting temperature distribution data from the initial laser parameter prediction results, calculating the temperature distribution using a finite element analysis tool, and obtaining characteristic data that matches stable cutting conditions; Step S73, using ANSYS software to calculate the thermal stress distribution based on the characteristic data, and determine a response value consistent with the characteristics of the brittle material; Step S74: if the response value exceeds a preset response value threshold, adjusting the laser power and scanning speed according to the difference between the response value and the response value threshold to obtain a corrected laser parameter set; Step S75: recalculating the temperature distribution and thermal stress distribution using a finite element analysis tool for the corrected laser parameter set to determine whether the stable cutting requirements are met; Step S76, based on the verification results, the training model is optimized using the TensorFlow framework to obtain the laser parameter output that matches the optimal combination; Step S77: Use the optimized laser parameter output to process the real-time dynamic data and determine the final laser parameter combination solution.

[0014] Optionally, in step S8, after obtaining the optimal laser parameter combination, the optimal laser parameter combination is applied in real time in the fully automatic laser cutting system, and the microcrack diffusion state is monitored by a sensor. If the microcrack diffusion area exceeds a preset microcrack diffusion area threshold, the auxiliary process optimization parameters are fine-tuned to obtain a final cutting quality control solution, including: Step S81, after obtaining the laser parameter data, the optimal combination is calculated using the least square method to obtain an initial laser parameter set; Step S82, loading the initial laser parameter set into the fully automatic cutting system, and generating cutting path data through real-time application; Step S83, collecting cutting path data through a sensor, and using an image processing algorithm to extract the microcrack diffusion state from the collected data to obtain a diffusion area value; Step S84, if the diffusion area value exceeds the microcrack diffusion area threshold, then adjusting the auxiliary process parameters by analyzing the diffusion state, and using an optimization algorithm to obtain an optimized process parameter set; Step S85, updating the cutting system with the optimized process parameter set, and generating adjusted cutting data through real-time application; Step S86, monitoring the adjusted cutting data through sensors, using image recognition algorithms to determine the microcrack diffusion state from the monitoring results, and obtaining final cutting quality parameters; Step S87: Adjust the laser parameters according to the final cutting quality parameters, and obtain a stable control solution through iterative calculation.

[0015] A second aspect of the present invention provides a thermal stress control system for laser cutting of brittle materials, which uses the above-mentioned method to control thermal stress in laser cutting of brittle materials. The system includes: The temperature distribution acquisition module is used to obtain real-time temperature distribution data of brittle materials during laser cutting. It uses infrared thermal imaging technology to capture the temperature changes on the material surface. Based on the influence of heat accumulation and steep temperature gradients, it calculates the temperature gradient value and obtains the initial thermal distribution characteristics. The stress concentration analysis module is used to extract local thermal stress concentration areas from the initial thermal distribution characteristics, simulate the thermal stress distribution under the thermal stress runaway state using the finite element analysis method, and determine the potential location and range of microcrack propagation and the thermal stress concentration coefficient by combining the dynamic data of material properties; The thermal stress modeling module is used to construct a mathematical model of the relationship between thermal stress and material fracture based on the thermal stress concentration coefficient and temperature gradient value. It introduces dynamic parameters of material properties and iteratively calculates the thermal stress runaway threshold through the lambda algorithm to obtain the thermal stress control boundary conditions. The laser parameter optimization module is used to control boundary conditions based on thermal stress, adjust the laser parameter matching scheme, obtain the laser power and pulse frequency combination that dynamically adapts to the material properties through the pre-established laser energy distribution database, and determine the optimized laser parameters; The temperature gradient control module is used to control the steepness of the temperature gradient by introducing synchronous cooling airflow during the cutting process through optimized laser parameters and combined with auxiliary process optimization technology. The airflow intensity is dynamically adjusted using an airflow pressure regulation algorithm to obtain a smooth temperature gradient distribution. The cutting condition determination module is used to extract the probability of thermal shock induction from the gentle temperature gradient distribution. If the probability of thermal shock induction exceeds the preset thermal shock induction probability threshold, the cutting speed adjustment parameter is adjusted to reduce the impact of heat accumulation, determine the degree of real-time fracture risk reduction, and determine stable cutting conditions; The parameter prediction training module is used to train the parameter prediction model based on stable cutting conditions and dynamic data of material properties. It uses a convolutional neural network to process historical data of temperature distribution, thermal stress distribution and laser parameters, and outputs the optimal laser parameter combination that matches the characteristics of brittle materials. The quality control execution module is used to obtain the optimal laser parameter combination and apply it in real time in the fully automatic laser cutting system. It monitors the microcrack diffusion status through sensors. If the microcrack diffusion area exceeds the preset microcrack diffusion area threshold, the auxiliary process optimization parameters are fine-tuned to obtain the final cutting quality control plan.

[0016] The technical solution provided by the embodiment of the present invention has the following beneficial effects: The present invention provides a method and system for controlling thermal stress in laser cutting of brittle materials. The method obtains real-time temperature distribution through infrared thermal imaging, combines finite element analysis with dynamic data of material properties, constructs a thermal stress and fracture relationship model, uses the lambda algorithm to iteratively calculate the thermal stress runaway threshold, determines the control boundary conditions, optimizes laser parameters based on these boundary conditions, introduces a synchronous cooling airflow, uses an airflow pressure regulation algorithm to dynamically control the temperature gradient, processes historical data through a convolutional neural network, predicts the optimal laser parameter combination, and monitors the microcrack diffusion state in real time during the cutting process, and fine-tunes auxiliary process parameters when necessary.

[0017] The present invention effectively solves the problem of fracture caused by runaway thermal stress in laser cutting of brittle materials, improves cutting quality and efficiency, and provides a new technical solution for precision machining of brittle materials. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 The present invention is a flow chart of a method for controlling thermal stress in laser cutting of brittle materials.

[0019] Figure 2 The figure is a schematic structural diagram of a thermal stress control system for laser cutting of brittle materials according to the present invention. DETAILED DESCRIPTION

[0020] The following will describe the technical solutions in the embodiments of the present invention in detail with reference to the accompanying drawings. The described embodiments are only a part of the embodiments of the present invention.

[0021] like Figure 1 As shown, the first aspect of the present invention provides a method for controlling thermal stress in laser cutting of brittle materials, the method comprising: S1, obtains real-time temperature distribution data of brittle materials during laser cutting, captures the surface temperature changes of the material through infrared thermal imaging technology, calculates the temperature gradient value based on the influence of heat accumulation and steep temperature gradient, and obtains the initial thermal distribution characteristics.

[0022] Optionally, this step also includes: Step S11 , collecting surface temperature data of the brittle material during laser cutting by infrared thermal imaging technology to obtain real-time temperature distribution.

[0023] Step S12: Using a two-dimensional convolution algorithm to process the real-time temperature distribution and extract the temperature change trend.

[0024] Step S13: Calculate the temperature gradient of the heat accumulation area based on the extracted temperature change trend to obtain a gradient value distribution.

[0025] In step S14, a principal component analysis method is used to obtain a characteristic vector of the initial thermal distribution for the gradient value distribution, and to determine the influence range of the thermal accumulation.

[0026] Step S15: If the temperature gradient in the feature vector exceeds the temperature gradient threshold, the initial thermal distribution is smoothed by a mean filter with a kernel size of 3x3 to obtain an optimized thermal distribution.

[0027] Step S16: extracting distribution features from the optimized thermal distribution to determine the thermal stress concentration area.

[0028] Step S17: classify the thermal stress concentration areas using a support vector machine algorithm, use radial basis function as a kernel function, train a classification model, and determine the potential crack location.

[0029] Specifically, by collecting surface temperature data of brittle materials during laser cutting through infrared thermal imaging technology, the temperature changes on the material surface can be captured in real time.

[0030] For example, when cutting glass, an infrared thermal imager can record the temperature of the laser-affected area at 50 frames per second. Assuming an initial temperature of 25°C, the local temperature rapidly rises to 500°C after the laser is applied, while the edge area gradually drops to 100°C, forming a distinct temperature distribution map. The core of this method lies in its non-contact and high sensitivity, which accurately reflects the dynamic propagation of heat and provides a reliable data foundation for subsequent analysis.

[0031] In one possible implementation, when a two-dimensional convolution algorithm is used to process real-time temperature distribution, the temperature data may be considered as a two-dimensional matrix.

[0032] For example, assuming the matrix size is 100x100, and the pixels correspond to temperature values, a 3x3 convolution kernel (such as a Gaussian kernel) can be used for sliding calculation to smooth out the noise and highlight the temperature change trend.

[0033] For example, the trend of the temperature in the central area slowly decreasing from 500°C to 450°C is extracted, while the sudden changes at the edge are weakened. This processing not only improves the interpretability of the data but also helps reduce the impact of external interference on the analysis.

[0034] Specifically, when calculating the temperature gradient of the heat accumulation area based on the extracted temperature change trend, it can be achieved through the temperature difference between adjacent pixels.

[0035] For example, the difference between a central region of 500°C and an adjacent region of 400°C indicates a gradient of 100°C / mm, while the difference between 100°C and 80°C at the edge is 20°C / mm. The distribution of gradient values ​​reflects the degree of heat concentration; larger gradients indicate a greater likelihood of thermal stress initiating cracks. This method provides a visual representation of how heat diffuses outward from the laser's point of application.

[0036] Preferably, when the principal component analysis method is used to obtain the characteristic vector of the initial thermal distribution, the temperature gradient data can be subjected to dimensionality reduction processing.

[0037] For example, assuming there are 10 primary gradient directions, principal component analysis can be used to extract the first three eigenvectors, representing the heat diffusion trends in the horizontal, vertical, and diagonal directions, respectively. If the horizontal eigenvector shows a gradient concentrated at 300°C / mm, this indicates significant heat accumulation in that direction. This analysis helps determine the heat-affected zone and provides a basis for subsequent optimization.

[0038] It should be noted that if the temperature gradient in the feature vector exceeds a preset temperature gradient threshold (such as 200°C / mm), the initial thermal distribution is smoothed by a 3x3 mean filter.

[0039] For example, in a region with a gradient of 250°C / mm, filtering results in a more uniform temperature distribution, with the local peak reduced to 180°C / mm. This optimized thermal distribution reduces noise interference, highlights the main heat accumulation areas, and effectively improves data stability.

[0040] In one embodiment, when extracting distribution features from the optimized thermal distribution, attention may be paid to temperature peaks and gradient concentration areas.

[0041] For example, the peak temperature is still 450°C, and the surrounding area with a gradient greater than 150°C / mm is marked as a thermal stress concentration area. This extraction method identifies potential risk points and provides key information for crack prediction.

[0042] For example, when using a support vector machine algorithm to classify areas of thermal stress concentration, radial basis functions are used as the kernel function, effectively handling nonlinear relationships. Suppose the training data consists of 100 samples, 50 of which are labeled "high crack risk" (gradient greater than 200°C / mm) and 50 are labeled "low risk." After model training, new data with a gradient of 220°C / mm is classified as high risk. This method improves the accuracy of crack prediction through machine learning and reduces the subjectivity of human judgment.

[0043] The above combination of technologies forms a complete process, from data collection to crack location, with rigorous logic and progressive steps. The output of each step supports the next, ultimately achieving precise control of thermal stress distribution during laser cutting of brittle materials. The advantages of this approach include improved processing quality, reduced crack defects, and increased production efficiency and material utilization.

[0044] Optionally, the step S17, classifying the thermal stress concentration areas by a support vector machine algorithm, using a radial basis function as a kernel function, training a classification model, and determining the potential crack location, further includes: Step S171 , collecting the temperature distribution of the thermal stress concentration area through thermal imaging data to obtain a preliminary area division result.

[0045] Step S172 : grouping the preliminary region division results using a K-means clustering algorithm to determine the boundary range where thermal stress is concentrated.

[0046] Step S173: extracting spatial feature data from the boundary range to obtain a feature vector of thermal stress distribution.

[0047] Step S174: classify the feature vectors using a support vector machine algorithm, using a radial basis function as a kernel function to train a classification model.

[0048] Step S175: If the confidence level output by the classification model exceeds the confidence threshold, a high-risk area for potential cracks is determined.

[0049] Step S176: For high-risk areas, a sliding window method is used to obtain the local temperature change trend and determine the crack propagation direction.

[0050] Step S177: extract dynamic features from the crack propagation direction and determine the priority order of crack development.

[0051] Specifically, when collecting the temperature distribution of the thermal stress concentration area through thermal imaging data and obtaining the preliminary area division result, it can be understood as using an infrared thermal imager to capture the temperature changes on the material surface.

[0052] For example, when laser cutting brittle materials, a thermal imager may show that the temperature in a certain area rises rapidly from 50 degrees to 200 degrees, indicating that there is significant heat accumulation here.

[0053] Preferably, this acquisition method requires high frame rate equipment to ensure real-time performance, thereby providing a reliable data basis for subsequent analysis.

[0054] In a possible implementation, a K-means clustering algorithm is used to group the preliminary region division results to determine the boundary range of the thermal stress concentration.

[0055] For example, assuming thermal imaging data contains multiple temperature points, they can be divided into three categories: low-temperature, medium-temperature, and high-temperature zones. The high-temperature zone is likely concentrated near the laser impact point, and its boundaries are automatically identified using a clustering algorithm. For example, points with temperatures above 180°C are classified as areas of concentrated thermal stress. This method effectively distinguishes different levels of thermal impact.

[0056] Specifically, when extracting spatial characteristic data from the boundary range and obtaining the characteristic vector of thermal stress distribution, it can be achieved by analyzing the spatial change rate of temperature.

[0057] For example, the temperature in a certain boundary area transitions from 150 degrees to 200 degrees, with a distance of only 2 mm, indicating that the gradient here is steeper.

[0058] The eigenvectors may contain information such as gradient values ​​and directions, and are used to characterize the spatial distribution of thermal stress. The eigenvectors are classified using a support vector machine algorithm, using radial basis functions as kernel functions. The process of training the classification model can be understood as a form of pattern recognition.

[0059] For example, after training the model based on historical data and inputting a new feature vector, the model can determine whether the area belongs to a "high heat stress zone."

[0060] In one embodiment, if the gradient value of a certain area reaches 50 degrees / mm, the model outputs an 85% probability that it belongs to a potential crack area, which provides a basis for subsequent decision-making.

[0061] It should be noted that if the confidence level of the classification model output exceeds a preset confidence threshold, such as 80%, a high-risk area for potential cracks is determined.

[0062] For example, if the confidence level for a particular cutting area reaches 90%, while the preset confidence threshold is 80%, this indicates significant thermal stress concentration in this area, possibly due to a sudden temperature change, causing thermal stress to exceed the limit within the material. This determination helps to identify key areas requiring attention in advance.

[0063] For high-risk areas, the sliding window method is used to obtain the local temperature change trend, and the direction of crack propagation can be determined through time series analysis.

[0064] For example, a 5x5 pixel window with sliding scanning revealed that the temperature in a certain direction dropped from 200°C to 120°C, with the trend pointing to the right, suggesting that cracks may extend in this direction. This method can dynamically track the evolution path of thermal stress.

[0065] In one embodiment, dynamic features are extracted from the crack propagation direction to determine the priority order of crack development, which may be based on the temperature change rate and the area of ​​the region.

[0066] For example, if the temperature drop rate in a certain direction reaches 30 degrees per second and the affected area is large, it will have a higher priority. This ranking helps optimize subsequent process adjustments and prioritize the crack development path with the greatest risk.

[0067] S2, extract the local thermal stress concentration area from the initial thermal distribution characteristics, use the finite element analysis method to simulate the thermal stress distribution under the thermal stress runaway state, combine the dynamic data of material properties, judge the potential location and range of microcrack propagation, and determine the thermal stress concentration coefficient.

[0068] Optionally, this step also includes: Step S21 , obtaining the temperature field of the local area from the thermal distribution data, generating thermal stress distribution data through finite element method simulation, and obtaining a preliminary thermal stress concentration area.

[0069] Step S22 : for the preliminary thermal stress concentration area, the material properties are updated in combination with dynamic data, and the thermal stress change trend under the thermal stress state is calculated using ANSYS software to determine the distribution of high thermal stress points.

[0070] In step S23 , the support vector machine algorithm is used to classify the distribution of high thermal stress points, and based on the classification results, potential locations of microcrack propagation are obtained to obtain a potential location set.

[0071] Step S24 , performing secondary simulation on the potential position set by using the finite element method, analyzing the stress field change in combination with the material properties, and determining the microcrack diffusion range.

[0072] Step S25 , obtaining the distribution of thermal stress values ​​within the diffusion range. If the thermal stress value exceeds a characteristic threshold in the material characteristic, it is marked as a high-risk area, and a high-risk area set is obtained.

[0073] Step S26: Calculate the local thermal stress concentration coefficient for the high-risk area set, verify the coefficient distribution through simulation analysis, and determine the final concentration coefficient.

[0074] Step S27 , extracting characteristic data from the final concentration coefficient, analyzing the correlation between thermal stress concentration and microcrack diffusion in combination with the thermal stress state, and obtaining a comprehensive evaluation result.

[0075] Specifically, when obtaining the temperature field of a local area from the thermal distribution data, the temperature data can be converted into a gridded calculation model through the finite element method.

[0076] For example, consider a localized area of ​​a brittle material divided into 1,000 mesh elements, with each element corresponding to a temperature value, such as 450°C in the center and 120°C at the edges. These temperature values ​​are then fed into a thermal stress calculation using a finite element method simulation, generating a preliminary thermal stress distribution map.

[0077] For example, the central area may show higher thermal stress values, while the edges may show lower values, preliminarily revealing areas of thermal stress concentration.

[0078] In one possible implementation, when updating material properties in conjunction with dynamic data, ANSYS software can be used to import temperature field data and adjust the thermal expansion coefficient and elastic modulus of the material.

[0079] Specifically, assuming the initial material properties are set at a room temperature of 25°C, but laser cutting causes the local temperature to rise to 500°C, the dynamically updated thermal expansion coefficient may change from 5×10^-6 / °C to 6×10^-6 / °C. Software calculations show that the thermal stress value in the central area may increase from 200MPa to 300MPa, while the edge remains around 50MPa, clearly showing the distribution trend of high thermal stress points.

[0080] It should be noted that when using the support vector machine algorithm to classify high thermal stress points, the thermal stress value and position coordinates can be used as input features.

[0081] For example, the training data includes 500 high thermal stress points, of which 200 are over 250 MPa and are marked as “high risk”.

[0082] In one embodiment, if the thermal stress value of the new data is 280 MPa, the algorithm classifies it as high risk and further infers that microcracks may spread outward from these points, ultimately forming a set of potential locations, such as the five grid points in the central area.

[0083] Preferably, the updated material properties may be loaded when performing a secondary simulation of the potential position set using the finite element method.

[0084] For example, assuming a microcrack with an initial length of 0.1 mm, simulations show that the stress field is concentrated at the crack tip, with the local thermal stress increasing from 300 MPa to 400 MPa. Combined with the influence of the temperature gradient, the crack may extend by 0.2 mm longitudinally and 0.15 mm transversely, thus determining the propagation range.

[0085] For example, after obtaining the distribution of thermal stress values ​​within the diffusion range, assuming that the material tensile strength threshold is 350 MPa, the 400 MPa in the central area exceeds the characteristic threshold and is marked as a high-risk area.

[0086] It is understandable that if the thermal stress in the edge area is only 80 MPa, no marking is required. This method effectively identifies the key locations for potential crack growth.

[0087] In one embodiment, when calculating the local thermal stress concentration factor, the thermal stress ratio between grid cells may be analyzed through simulation.

[0088] For example, the thermal stress at the crack tip is 400 MPa, and that in the adjacent area is 100 MPa, with a concentration coefficient of 4. After verification, the coefficient distribution reflects the spatial characteristics of thermal stress concentration, providing a basis for subsequent analysis.

[0089] Specifically, when extracting feature data from the final concentrated coefficient, the area where the coefficient is greater than 3 can be focused on.

[0090] For example, a coefficient of 4 for the center grid indicates a high correlation between thermal stress concentration and microcrack propagation. Combined with thermal stress state analysis, this correlation helps pinpoint crack risk points and improves process control.

[0091] S3, based on the thermal stress concentration coefficient and temperature gradient value, constructs a mathematical model of the relationship between thermal stress and material fracture, introduces dynamic parameters of material properties, and iteratively calculates the thermal stress runaway threshold through the Lambda algorithm to obtain the thermal stress control boundary conditions.

[0092] Optionally, this step also includes: Step S31 : Based on the thermal stress concentration and temperature gradient data, a mathematical model is constructed using the finite element method to calculate the thermal stress value distribution.

[0093] Step S32: Input the thermal stress value distribution result into the Laplace iterative algorithm for iterative calculation to determine the dynamic parameter change trend.

[0094] Step S33: adjusting the thermal stress value using the material constitutive relationship according to the changing trend of the dynamic parameters.

[0095] Step S34: If the adjusted thermal stress value exceeds the preset thermal stress out-of-control threshold, the control boundary is recalculated using a mathematical model.

[0096] Step S35 , limiting the thermal stress value by regulating the boundary, and using the fracture mechanics criterion to determine the conditions for material fracture.

[0097] Step S36: obtaining the concentrated distribution of thermal stress under the material fracture condition, and iteratively updating the thermal stress out-of-control threshold using the least squares method.

[0098] Step S37: Using the updated thermal stress runaway threshold, determine the final control boundary condition.

[0099] Specifically, based on the thermal stress concentration and temperature gradient data, the temperature field and stress field can be used as the input basis when constructing a mathematical model using the finite element method.

[0100] For example, suppose a metal plate is laser processed, the temperature in the center area reaches 600°C and the edge is 150°C. The finite element method is used to divide the area into 2000 grid cells, each cell corresponds to a temperature value and an initial thermal stress value, forming a continuous distribution model.

[0101] In one possible implementation, software such as Abaqus can import this data and, combined with meshing, calculate thermal stress values ​​of approximately 320 MPa in the center, dropping to 60 MPa at the edges, providing a preliminary understanding of the thermal stress distribution. Inputting this thermal stress distribution into the Laplace iterative algorithm can iteratively smooth the thermal stress trend.

[0102] Specifically, the initial thermal stress distribution may show abrupt changes due to the temperature gradient. The Laplace algorithm gradually adjusts the thermal stress difference between grids through multiple iterations.

[0103] For example, the mutation between 320 MPa in the central area and 200 MPa in the adjacent area is smoothed to 280 MPa and 220 MPa after 5 iterations, revealing the dynamic change trend of thermal stress.

[0104] It should be noted that this method can effectively capture the law of thermal stress changes with time or temperature. When adjusting the thermal stress value according to the trend of dynamic parameter changes, the material constitutive relationship can be introduced.

[0105] In one embodiment, assuming that the elastic modulus of the material decreases from 200 GPa to 180 GPa at high temperature, combined with the thermal stress distribution trend, the thermal stress value of the central area is adjusted from 280 MPa to 260 MPa.

[0106] Preferably, this adjustment takes into account the actual response characteristics of the material with temperature changes, ensuring that the thermal stress value is closer to reality. If the adjusted thermal stress value exceeds the preset thermal stress runaway threshold, for example, set to 300MPa, the control boundary needs to be recalculated.

[0107] It is understandable that the thermal stress value of 260 MPa in the central area does not exceed the standard, but if a local point reaches 310 MPa due to abnormal temperature rise, the boundary conditions can be adjusted through mathematical models, such as reducing the input heat or changing the cooling rate, so that the thermal stress can be brought back to a safe range.

[0108] After limiting the thermal stress value by regulating the boundary, the fracture conditions are determined using the fracture mechanics criteria.

[0109] For example, assuming the fracture toughness of the material is 20 MPa·m^0.5, combined with the thermal stress distribution, if the thermal stress concentration at a certain point causes the equivalent stress intensity factor to exceed this value, it is considered that fracture may occur.

[0110] In one embodiment, the thermal stress in the central area is 260 MPa, and the crack tip factor is calculated to be within the limit, so it is determined that there is no risk of fracture.

[0111] After obtaining the concentrated distribution of thermal stress under fracture conditions, the least squares method can be used to iteratively update the thermal stress runaway threshold.

[0112] Specifically, if the initial thermal stress runaway threshold of 300 MPa is too conservative, the thermal stress runaway threshold can be updated to 310 MPa by analyzing multiple simulation data to make it more in line with actual working conditions.

[0113] For example, this iteration can optimize the applicability of the threshold. When the updated thermal stress runaway threshold is used to determine the final control boundary conditions, for example, after setting the thermal stress runaway threshold to 310 MPa, re-simulation found that the thermal stress in the central area is stable at 270 MPa. Through multi-faceted verification, such as temperature control, thermal stress distribution smoothness, and material property adjustment, the boundary conditions are ensured to be both strict and reasonable. The gradual advancement of this method will help improve the accuracy and reliability of thermal stress management.

[0114] S4, according to the boundary conditions of thermal stress regulation, adjust the laser parameter matching scheme, obtain the laser power and pulse frequency combination that dynamically adapts to the material properties through the pre-established laser energy distribution database, and determine the optimized laser parameters.

[0115] Optionally, this step also includes: In step S41 , initial values ​​of laser parameters related to boundary conditions are obtained through a pre-established laser energy distribution database to obtain a preliminary matching solution.

[0116] Step S42: In the preliminary matching solution, material characteristic data is introduced, dynamic adaptation characteristics are extracted, and the corresponding laser power range is determined.

[0117] Step S43: After determining the laser power range, a binary search algorithm is used to adjust the pulse frequency to obtain a frequency combination that matches the laser power.

[0118] Step S44: If the thermal stress control requirement changes, the laser parameters are re-matched and the matching solution is updated through the energy distribution data in the database.

[0119] Step S45 : According to the updated matching scheme, laser parameters matching the boundary conditions are obtained, and an adjusted thermal stress control scheme is determined.

[0120] Step S46: In the thermal stress control scheme, linear regression is used to analyze the correlation between material properties and laser power to determine the parameter optimization direction.

[0121] Step S47: According to the parameter optimization direction, the ratio of pulse frequency to laser power is adjusted to determine the final laser parameters.

[0122] Specifically, when obtaining initial values ​​of laser parameters related to boundary conditions through a pre-established laser energy distribution database, data matching a specific processing scenario can be extracted from the database.

[0123] For example, let's say a 5mm thick steel plate is being laser processed. The database records the initial production parameters corresponding to different energy distributions, such as a laser power of 1000 watts and a pulse frequency of 500 Hz. These initial laser parameter values ​​can provide a benchmark for subsequent adjustments.

[0124] In one possible implementation, the database might contain parameter combinations for various boundary conditions, such as edge cooling or high-temperature core scenarios, allowing users to directly access the data that best matches the current operating conditions. When extracting dynamic adaptive features using material property data, parameters can be adjusted based on the material's thermal conductivity and expansion coefficient.

[0125] Specifically, the thermal conductivity of steel plates may drop from 50 W / m·°C to 45 W / m·°C at high temperatures. By analyzing this change trend, the laser power range is determined to be between 800 W and 1200 W.

[0126] It should be noted that this range reflects the material's response to heat input, ensuring that thermal stresses during processing are controllable. When adjusting the pulse frequency using a binary search algorithm, the frequency range can be set from 200 Hz to 800 Hz.

[0127] For example, if the initial frequency of 500 Hz corresponds to high thermal stress, a binary approach is used to test 400 Hz and 600 Hz, gradually converging to 450 Hz, which matches 1000 watts of power. This method quickly finds the optimal combination and improves parameter adjustment efficiency. If the thermal stress control requirements change and the laser parameters need to be re-adjusted, the energy distribution data in the database can be used to update the solution.

[0128] In one embodiment, if an increase in cooling rate results in thermal stress concentration, a low-power, high-frequency combination, such as 800 watts and 700 Hz, can be extracted from the database to quickly adapt to the new demand.

[0129] Preferably, such an update can maintain processing stability. When obtaining laser parameters that match the boundary conditions, specific values ​​can be determined based on the adjusted solution.

[0130] For example, setting the power to 900 watts and the frequency to 600 Hz ensures uniform thermal stress distribution. This parameter combination effectively addresses dynamic changes in boundary conditions. When using linear regression to analyze the correlation between material properties and laser power, trends can be identified across multiple sets of experimental data.

[0131] For example, if the power increases from 800 watts to 1000 watts, the surface temperature of the material increases by 20%. The regression analysis can indicate the positive impact of the power increase on the thermal stress and guide the optimization direction.

[0132] In one embodiment, this analysis can also reveal the priority of frequency adjustments.

[0133] When adjusting the ratio of pulse frequency to laser power, the parameters can be fine-tuned according to the optimization direction.

[0134] Specifically, if regression analysis shows that frequency has a greater impact on thermal stress, the frequency can be adjusted from 600 Hz to 650 Hz, while maintaining the power at 900 watts, ultimately determining the actual parameters used. This ratio adjustment can refine thermal stress management and improve processing accuracy and material integrity.

[0135] It is understandable that verifying the rationality of parameters from multiple aspects helps to form a reliable solution.

[0136] Optionally, step S46, in the thermal stress control scheme, uses linear regression to analyze the correlation between material properties and laser power to determine the parameter optimization direction, and further includes: Step S461 : Acquire characteristic data related to thermal stress regulation through a pre-established material property database to determine an initial analysis range.

[0137] In step S462 , linear regression is used to analyze the correlation between material characteristics and laser power within the initial analysis range to obtain a characteristic change trend.

[0138] Step S463: extract the production parameter set that matches the control scheme from the energy distribution data according to the characteristic change trend, and determine the production parameter screening direction.

[0139] Step S464: If the production parameter screening direction deviates from expectations, the material characteristic data is updated through the database to obtain an adjusted feature set.

[0140] Step S465: cluster analysis is used to classify the priorities of production parameter adjustments based on the adjusted feature set, and to determine the optimized production parameter combination.

[0141] Step S466: By optimizing the production parameter combination, the boundary conditions of the control scheme are adjusted to obtain a balanced scheme for thermal stress distribution.

[0142] Step S467: According to the balancing solution, the final production parameter ratio is extracted from the matching solution to determine the execution parameters of the thermal stress control.

[0143] Specifically, a pre-established material property database is used to obtain characteristic data related to thermal stress regulation and determine the initial analysis range. This approach relies on material information stored in the database, such as the thermal stress response characteristics of metals or alloys at different temperatures.

[0144] For example, for a high-strength steel, the database might record a yield strength of 800 MPa at 200°C, dropping to 600 MPa at 500°C. This data provides a baseline range for subsequent analysis, helping to identify a starting point for thermal stress control. Within this initial analysis range, linear regression is used to analyze the correlation between material properties and laser power, identifying characteristic variation trends.

[0145] It is understood that linear regression can reveal how laser power affects material temperature and thermal stress distribution.

[0146] In one possible implementation, experimental data shows that increasing laser power from 1000W to 1500W increases the material surface temperature by 150°C and reduces the distribution of concentrated thermal stress areas by 20%. This trend provides guidance for adjusting production parameters. Based on the characteristic change trend, the energy distribution data is used to extract a production parameter set that matches the control solution, thus determining the direction for production parameter selection.

[0147] Specifically, energy distribution data might show that at 1200W the heat input is concentrated at a depth of 2mm, while at 1800W it extends to 5mm.

[0148] Preferably, if the goal is shallow thermal stress control, low-power production parameter sets are prioritized. This screening ensures that the solution matches actual needs. If the production parameter screening direction deviates from expectations, the material property data is updated through the database to obtain an adjusted feature set.

[0149] For example, when the ambient humidity increases from 30% to 70%, the thermal conductivity of the material may decrease by 5%, requiring the updated data to be reloaded.

[0150] In one embodiment, for aluminum alloy, it is found that its thermal expansion coefficient is 23×10⁻ 6 / °C adjusted to 24×10⁻ 6 / °C. This adjustment improves the adaptability of production parameters. Based on the adjusted feature set, cluster analysis is used to prioritize production parameter adjustments and determine the optimal production parameter combination.

[0151] It should be noted that cluster analysis can classify production parameters into high-priority groups and low-priority groups.

[0152] For example, for stainless steel, the clustering results may show that the combination of 1300W power and 50Hz frequency is more suitable for fast regulation, while 1500W and 30Hz are more suitable for stable regulation. This division optimizes resource allocation.

[0153] By optimizing the combination of production parameters and adjusting the boundary conditions of the control scheme, a balanced solution for thermal stress distribution is obtained.

[0154] In a possible implementation, after the boundary conditions are adjusted, the thermal stress changes from being concentrated at the edge to being evenly distributed.

[0155] For example, in the initial design, thermal stress at the edge reached 500 MPa, while at the center it was only 200 MPa. After adjustment, both stabilized at around 350 MPa. This balance improved the material's service life. Based on the balanced design, the final production parameter ratio was extracted from the matching plan to determine the execution parameters for thermal stress control.

[0156] For example, after multiple rounds of analysis, 1400W power and 40Hz frequency were ultimately selected as the execution parameters. This combination demonstrated good stability and adaptability in the experiment, providing a reliable basis for subsequent processing.

[0157] Optionally, the execution parameters can be used as optimized laser parameters in subsequent steps.

[0158] S5, through optimized laser parameters combined with auxiliary process optimization technology, introduces synchronous cooling airflow during the cutting process to control the steepness of the temperature gradient, and uses the airflow pressure adjustment algorithm to dynamically adjust the airflow intensity to obtain a smooth temperature gradient distribution.

[0159] Optionally, this step also includes: Step S51 : generating initial cutting data by adjusting laser parameters, and using a preset temperature gradient change trend threshold to judge the temperature gradient change trend, to obtain preliminary temperature distribution information.

[0160] Step S52: obtaining cooling airflow requirements based on the temperature distribution information, adjusting the airflow output using synchronous control technology, and determining an initial value of the airflow intensity.

[0161] In step S53, an airflow pressure adjustment algorithm is introduced based on the initial value of the airflow intensity to dynamically adjust the airflow pressure to obtain a stable airflow control parameter.

[0162] Step S54 , by means of stable airflow control parameters and combined with auxiliary process technology, the temperature gradient during the cutting process is adjusted to obtain a gentle temperature distribution characteristic.

[0163] Step S55: If the temperature distribution characteristics exceed the preset range, the airflow intensity is recalculated through the adjustment algorithm to obtain optimized airflow parameters.

[0164] Step S56: Adjust the synchronization control logic according to the optimized airflow parameters, determine the matching degree between the airflow and laser parameters, and obtain the final process optimization data.

[0165] Step S57: Generate a cutting result with a gentle temperature gradient through the final process optimization data to determine the achievement status of the business goal.

[0166] Specifically, when generating initial cutting data through optimized laser parameter adjustment, the temperature field during the processing can be simulated based on the preset laser power and pulse frequency.

[0167] For example, let's assume we're processing a 5mm thick steel plate with a laser power of 1000W, a pulse frequency of 500Hz, and a temperature gradient threshold of 20°C / mm. Using a heat conduction model to determine the temperature trend, we can obtain preliminary temperature distribution information. This distribution information provides foundational data for subsequent steps.

[0168] In a possible implementation, when obtaining the cooling airflow requirement based on the temperature distribution information, the peak temperature of the steel plate surface may be analyzed.

[0169] For example, if the temperature distribution shows a temperature of 600°C in the center and 300°C at the edges, cooling airflow will be needed to reduce the gradient. Synchronous control technology is used to adjust the airflow output, with an initial setting of 10 liters / minute. This is dynamically adjusted based on the temperature difference, ensuring that airflow and processing are synchronized.

[0170] It should be noted that when the airflow pressure adjustment algorithm is introduced according to the initial value of the airflow intensity, the pressure can be adjusted by real-time monitoring of temperature feedback.

[0171] Specifically, if the initial airflow pressure is 2 bar but the temperature gradient remains high, the algorithm can gradually increase the pressure to 2.5 bar to achieve stable airflow control parameters. This dynamic adjustment allows for rapid response to processing requirements. When adjusting the temperature gradient through stable airflow control parameters combined with auxiliary process technologies, auxiliary cooling nozzles can be introduced.

[0172] For example, by injecting cooling air synchronously along the laser cutting path, the temperature distribution is smoothed from a steep 50 degrees per millimeter to 15 degrees per millimeter. This smooth feature helps reduce thermal deformation.

[0173] Preferably, if the temperature distribution characteristic exceeds a preset range, for example, the gradient exceeds 25 degrees / mm, the airflow intensity is recalculated through an adjustment algorithm.

[0174] In one embodiment, the air flow can be increased from 10 L / min to 15 L / min, combined with the pressure being adjusted to 3 bar, and the optimized parameters can effectively flatten the temperature curve.

[0175] It is understandable that when adjusting the synchronization control logic according to the optimized airflow parameters, it is necessary to determine the degree of matching between the airflow and laser parameters.

[0176] For example, when 900 watts of power is combined with a 600 Hz frequency, the airflow intensity is adjusted to 12 liters / minute. By comparing the temperature distribution, the matching is verified and the final process optimization data is generated. This matching can improve processing consistency.

[0177] In one embodiment, when the cutting result is generated by the final process optimization data, it can be observed that the temperature gradient on the surface of the steel plate is stable within 10 degrees / mm.

[0178] For example, the smoothness of the cut edge was improved, and the width of the heat-affected zone was reduced from 1 mm to 0.5 mm, achieving the business goal. This result verifies the practicality of parameter optimization.

[0179] S6, extracting the probability of thermal shock induction from the gentle temperature gradient distribution. If the probability of thermal shock induction exceeds the preset threshold value of thermal shock induction probability, the cutting speed adjustment parameter is adjusted to reduce the influence of heat accumulation, judge the degree of reduction of real-time fracture risk, and determine the stable cutting conditions.

[0180] Optionally, this step also includes: Step S61 : acquiring temperature gradient distribution data through a sensor, and calculating the probability of thermal shock induction using kernel density estimation to obtain a probability value.

[0181] Step S62: If the probability value exceeds a preset thermal shock induction probability threshold, the relationship between thermal shock and heat accumulation is predicted by linear regression to determine the cutting speed adjustment range.

[0182] Step S63: updating the cutting speed parameter according to the adjustment range, using a real-time monitoring tool to obtain the trend of heat accumulation changes, and determining the degree of heat accumulation reduction.

[0183] Step S64 , calculating the fracture risk change rate based on the relationship between the degree of heat accumulation reduction and the fracture risk, and obtaining a risk reduction index.

[0184] Step S65: If the risk reduction index reaches the stable threshold, the stable cutting state is confirmed by the ARIMA model to determine the validity of the current cutting speed parameter.

[0185] Step S66: Acquire real-time fracture risk data, compare the risk reduction index with the stability threshold, and determine whether the cutting speed parameters need to be further adjusted.

[0186] Step S67, by iterating the above process in a loop, the temperature gradient data is updated using kernel density estimation to obtain the optimized stable cutting conditions.

[0187] Specifically, when obtaining temperature gradient distribution data through a sensor, an infrared thermometer can be used to scan the surface temperature of the processing area in real time.

[0188] For example, when cutting a 5mm thick steel plate, the sensor collects 100 temperature points per second, covering the area from the center of the cut to the edge, generating a continuous temperature gradient distribution map. When using kernel density estimation to calculate the probability of thermal shock induction, the collected temperature data can be used to analyze the central trend of the gradient change.

[0189] For example, if the temperature gradient fluctuates between 20°C / mm and 30°C / mm, kernel density estimation can yield a thermal shock probability of 0.3. This probability reflects the likelihood of material cracking due to a sudden temperature change, providing a basis for subsequent decision-making.

[0190] In a possible implementation, if the heat shock probability exceeds a preset heat shock induction probability threshold of 0.25, the relationship between heat shock and heat accumulation is predicted by linear regression.

[0191] Specifically, the model can be trained using historical processing data, with input variables including temperature gradient peak and processing time, and output being heat accumulation.

[0192] For example, when the temperature gradient is 28 degrees / mm and the cutting duration is 10 seconds, the predicted heat accumulation is 500 joules. Therefore, it is determined that the cutting speed needs to be reduced by 10% to reduce heat input. When the cutting speed parameter is updated based on the adjustment range, it can be reduced from the initial 2 mm / s to 1.8 mm / s. Real-time monitoring tools such as thermocouples can track the trend of heat accumulation. For example, if heat accumulation decreases from 500 joules to 450 joules, it indicates that heat dissipation has improved.

[0193] It should be noted that when calculating the fracture risk change rate by the degree of heat accumulation reduction, the relationship between the heat accumulation reduction and the thermal stress release of the material can be analyzed.

[0194] In one embodiment, a 50-joule reduction in heat accumulation may reduce the fracture risk from 0.2 to 0.15, a change rate of -0.05. This risk reduction indicator reflects the effectiveness of the cutting speed parameter adjustment. If the indicator reaches a preset stability threshold of 0.1, the stability of the cutting state is confirmed through the ARIMA model.

[0195] For example, based on the temperature and risk data of the past 20 seconds, the model predicts that the risk fluctuation in the next 5 seconds will be less than 0.02, proving that the current cutting speed parameters are reasonable.

[0196] Specifically, when obtaining real-time fracture risk data, an ultrasonic detector can be used to monitor changes in microcracks inside the material.

[0197] For example, if the crack length detected at the cutting edge decreases from 0.2 mm to 0.1 mm, and the risk reduction indicator matches the stability threshold of 0.1, no further parameter adjustment is required.

[0198] Preferably, when the temperature gradient data is updated through cyclic iteration, the temperature distribution can be re-collected in the next round of cutting.

[0199] For example, the adjusted gradient stabilized at 15 degrees / mm, and kernel density estimation showed that the probability of thermal shock dropped below 0.2. This continuous optimization ensured the stability of processing conditions.

[0200] It can be understood that this method gradually approaches the optimal cutting state through real-time data feedback and dynamic adjustment.

[0201] For example, when the initial speed is too fast, resulting in excessive heat accumulation, reducing the speed and monitoring the effect can effectively control the risk.

[0202] In one embodiment, after multiple iterations, the temperature gradient was optimized from 25°C / mm to 12°C / mm, significantly reducing the risk of fracture. This strategy improves processing reliability and consistency, providing data support for subsequent process optimization.

[0203] S7, based on the stable cutting conditions and dynamic data of material properties, trains the parameter prediction model, uses a convolutional neural network (CNN) to process the historical data of temperature distribution, thermal stress distribution and laser parameters, and outputs the optimal laser parameter combination that matches the characteristics of brittle materials.

[0204] Optionally, this step also includes: Step S71: Process historical data through a convolutional neural network, input the historical data as time series temperature and thermal stress data, use the TensorFlow framework for training, and obtain the initial laser parameter prediction results.

[0205] Step S72: extracting temperature distribution data from the initial laser parameter prediction results, calculating the temperature distribution using a finite element analysis tool, and obtaining characteristic data that matches the stable cutting conditions.

[0206] Step S73: Calculate the thermal stress distribution using ANSYS software based on the characteristic data to determine a response value consistent with the characteristics of the brittle material.

[0207] Step S74 : If the response value exceeds a preset response value threshold, the laser power and scanning speed are adjusted according to the difference between the response value and the response value threshold to obtain a corrected laser parameter set.

[0208] Step S75 : recalculating the temperature distribution and thermal stress distribution using a finite element analysis tool for the corrected laser parameter set to determine whether the stable cutting requirement is met.

[0209] Step S76: By verifying the results, the training model is optimized using the TensorFlow framework to obtain the laser parameter output that matches the optimal combination.

[0210] Step S77: Use the optimized laser parameter output to process the real-time dynamic data and determine the final laser parameter combination solution.

[0211] Specifically, when processing historical data through convolutional neural networks, time series temperature and thermal stress data can be used for analysis.

[0212] For example, when cutting a 5mm thick steel plate, the temperature and thermal stress values ​​were collected every second for the past 30 minutes, forming a sequence of 1,800 data points. The core of this method is to capture the cyclical trend in the data.

[0213] One possible implementation involves building a network using the TensorFlow framework. The input layer receives temperature and thermal stress sequences, while the convolutional layer extracts local features and ultimately outputs the initial laser parameters. This approach can quickly adapt to data patterns under varying processing conditions.

[0214] When extracting temperature distribution data from the initial laser parameters, it can be further processed in conjunction with finite element analysis tools.

[0215] Specifically, assuming the prediction results show that the temperature in the center of the cutting area is 600 degrees and the edge is 200 degrees, this data can be imported into the tool to generate a temperature field distribution map.

[0216] In one embodiment, the steel plate is divided into 1,000 cells through meshing, and the temperature value of each cell is calculated. This distribution provides intuitive basic data for subsequent analysis. When using ANSYS software to calculate the thermal stress distribution, the material response can be further deduced based on the temperature distribution.

[0217] It should be noted that thermal stress concentration of brittle materials usually occurs in areas with large temperature gradients.

[0218] For example, at a center temperature of 600°C and an edge temperature of 200°C, the software calculates that the maximum thermal stress could reach 300 MPa. This thermal stress distribution reflects potential risk points during material processing and provides a basis for adjusting laser parameters. If the response value exceeds the preset threshold, for example, if the thermal stress threshold is set at 250 MPa and the calculated value is 300 MPa, the laser power and scanning speed need to be adjusted.

[0219] In one possible implementation, based on the difference of 50 MPa, the laser power is reduced from 1000 W to 900 W, while the scanning speed is adjusted from 2 mm / s to 1.8 mm / s. This adjustment relieves thermal stress concentration by reducing heat input.

[0220] When recalculating for the modified laser parameter set, the effect can be verified using finite element analysis tools.

[0221] For example, after adjustment, the core temperature dropped to 550 degrees Celsius and the maximum thermal stress was reduced to 240 MPa. This change shows that the laser parameter correction effectively reduced the risk.

[0222] Preferably, multiple iterations of the calculation can be performed to ensure that the temperature and thermal stress distribution meet the requirements for stable cutting. When the training model is optimized through verification results, the laser parameters can be updated using the TensorFlow framework.

[0223] For example, by re-inputting adjusted temperature and thermal stress data into the model, the model outputs optimal laser parameters after training, such as 920 watts of power and 1.9 mm / s of speed. This optimization improves the model's adaptability to real-time data. When processing real-time dynamic data using the optimized laser parameters, the final solution can be determined in conjunction with the latest temperature and thermal stress values ​​collected by the sensor.

[0224] In one embodiment, real-time monitoring shows that the temperature is stable at 540 degrees and the thermal stress is kept below 230 MPa. This dynamic adjustment ensures the stability and consistency of the processing.

[0225] It is understandable that this method can effectively deal with uncertainties in processing and improve overall efficiency through data-driven and real-time feedback.

[0226] S8, after obtaining the optimal laser parameter combination, is applied in real time in the fully automatic laser cutting system. The microcrack diffusion state is monitored by sensors. If the microcrack diffusion area exceeds the preset microcrack diffusion area threshold, the auxiliary process optimization parameters are fine-tuned to obtain the final cutting quality control solution.

[0227] Optionally, this step also includes: Step S81: After acquiring the laser parameter data, the optimal combination is calculated using the least square method to obtain an initial laser parameter set.

[0228] Step S82: Load the initial laser parameter set into the fully automatic cutting system and generate cutting path data through real-time application.

[0229] Step S83: collecting cutting path data through sensors, and using image processing algorithms to extract the microcrack diffusion state from the collected data to obtain a diffusion area value.

[0230] Step S84: If the diffusion area value exceeds the preset microcrack diffusion area threshold, the auxiliary process parameters are adjusted by analyzing the diffusion state, and an optimized process parameter group is obtained using an optimization algorithm.

[0231] Step S85 , using the optimized process parameter set to update the cutting system, and generating adjusted cutting data through real-time application.

[0232] Step S86: Monitor the adjusted cutting data through sensors, use image recognition algorithms to determine the microcrack diffusion state from the monitoring results, and obtain the final cutting quality parameters.

[0233] Step S87: Adjust the laser parameters according to the final cutting quality parameters, and obtain a stable control solution through iterative calculation.

[0234] Specifically, after obtaining the laser parameter data, the least square method can be used to calculate the optimal combination.

[0235] For example, when cutting a steel plate with a thickness of 5 mm, it is assumed that the collected data includes multiple parameters such as laser power, scanning speed, and focus position.

[0236] In one possible implementation, historical cutting data is first organized into a matrix format, which contains power values ​​from multiple experiments, such as 800 watts to 1200 watts, and speed values, such as 1.5 mm / s to 2.5 mm / s. These data are fitted using the least squares method to obtain an initial parameter set, such as 1000 watts of power and 2 mm / s of speed. The core of this method is to quickly lock in the parameter range through mathematical optimization. After loading the initial parameters in the fully automatic cutting system, cutting path data needs to be generated.

[0237] Specifically, the system's built-in path planning module can be used to convert the parameter group into a specific motion trajectory.

[0238] For example, for a rectangular steel plate, the system generates a straight path from the lower left corner to the upper right corner according to the parameter group, and the path length is 500 mm.

[0239] It should be noted that this path data directly affects the accuracy and efficiency of subsequent cutting. When the cutting path data is collected by the sensor, the image processing algorithm can be used to analyze the microcrack diffusion state.

[0240] In one embodiment, a sensor captures high-definition images of the cutting area every second, with a resolution of 1920 x 1080 pixels. An algorithm uses edge detection to identify crack boundaries and calculates the crack's propagation area. For example, if a crack's propagation area is found to be 2 square millimeters, this method can intuitively reflect the extent of material damage.

[0241] If the diffusion area value exceeds a preset microcrack diffusion area threshold, for example, if the microcrack diffusion area threshold is set to 1.5 square millimeters, the auxiliary process parameters need to be adjusted.

[0242] Preferably, the direction and speed of crack propagation can be analyzed and an optimization algorithm such as a genetic algorithm can be used to adjust the air pressure or cooling water flow rate.

[0243] For example, the initial pressure is 0.5 MPa, which is then increased to 0.6 MPa after adjustment. This optimized parameter set can effectively suppress crack propagation. After updating the system with the optimized parameter set, the adjusted cutting data is generated.

[0244] It is understandable that the system will re-plan the path and perform the cutting according to the new parameters.

[0245] For example, after adjustment, the path remains unchanged, but the cutting speed is slightly reduced to 1.8 mm / s. This adjustment improves process stability by reducing heat buildup. When the sensor monitors the adjusted data, the image recognition algorithm can again be used to determine the crack status.

[0246] In one possible implementation, monitoring showed that the crack area was reduced to 1.2 square millimeters, indicating that the adjustment was effective. This real-time feedback provides a reliable basis for quality control.

[0247] When adjusting the laser parameters according to the final cutting quality parameters, a stable solution can be determined through iterative calculation.

[0248] For example, if the quality parameters show that the edge roughness is still too high, the power can be fine-tuned from 1000 W to 980 W while the speed remains stable at 1.9 mm / s.

[0249] In one embodiment, after three iterations, the crack area stabilized at less than 1 square millimeter. This solution ensures the efficiency and consistency of the cutting process through continuous optimization.

[0250] like Figure 2 As shown, the second aspect of the present invention provides a thermal stress control system for laser cutting of brittle materials, which uses the above-mentioned method to control thermal stress in laser cutting of brittle materials. The system includes: The temperature distribution acquisition module is used to obtain real-time temperature distribution data of brittle materials during laser cutting. It uses infrared thermal imaging technology to capture the temperature changes on the material surface. Based on the influence of heat accumulation and steep temperature gradients, it calculates the temperature gradient value and obtains the initial thermal distribution characteristics. The stress concentration analysis module is used to extract local thermal stress concentration areas from the initial thermal distribution characteristics, simulate the thermal stress distribution under the thermal stress runaway state using the finite element analysis method, and determine the potential location and range of microcrack propagation and the thermal stress concentration coefficient by combining the dynamic data of material properties; The thermal stress modeling module is used to construct a mathematical model of the relationship between thermal stress and material fracture based on the thermal stress concentration coefficient and temperature gradient value. It introduces dynamic parameters of material properties and iteratively calculates the thermal stress runaway threshold through the Lambda algorithm to obtain the thermal stress control boundary conditions. The laser parameter optimization module is used to control boundary conditions based on thermal stress, adjust the laser parameter matching scheme, obtain the laser power and pulse frequency combination that dynamically adapts to the material properties through the pre-established laser energy distribution database, and determine the optimized laser parameters; The temperature gradient control module is used to control the steepness of the temperature gradient by introducing synchronous cooling airflow during the cutting process through optimized laser parameters combined with auxiliary process optimization technology. The airflow pressure adjustment algorithm is used to dynamically adjust the airflow intensity to obtain a smooth temperature gradient distribution. The cutting condition determination module is used to extract the probability of thermal shock induction from the gentle temperature gradient distribution. If the probability of thermal shock induction exceeds the preset thermal shock induction probability threshold, the cutting speed adjustment parameter is adjusted to reduce the impact of heat accumulation, determine the degree of real-time fracture risk reduction, and determine stable cutting conditions; The parameter prediction training module is used to train the parameter prediction model based on stable cutting conditions and dynamic data of material properties. It uses a convolutional neural network (CNN) to process historical data of temperature distribution, thermal stress distribution, and laser parameters, and outputs the optimal laser parameter combination that matches the characteristics of the brittle material. The quality control execution module is used to obtain the optimal laser parameter combination and apply it in real time in the fully automatic laser cutting system. It monitors the microcrack diffusion status through sensors. If the microcrack diffusion area exceeds the preset microcrack diffusion area threshold, the auxiliary process optimization parameters are fine-tuned to obtain the final cutting quality control plan.

[0251] The present invention provides a method and system for controlling thermal stress in laser cutting of brittle materials. The method obtains real-time temperature distribution through infrared thermal imaging, combines finite element analysis with dynamic data of material properties, constructs a thermal stress and fracture relationship model, uses the lambda algorithm to iteratively calculate the thermal stress runaway threshold, determines the control boundary conditions, optimizes laser parameters based on these boundary conditions, introduces a synchronous cooling airflow, uses an airflow pressure regulation algorithm to dynamically control the temperature gradient, processes historical data through a convolutional neural network, predicts the optimal laser parameter combination, and monitors the microcrack diffusion state in real time during the cutting process, and fine-tunes auxiliary process parameters when necessary.

[0252] The present invention effectively solves the problem of fracture caused by runaway thermal stress in laser cutting of brittle materials, improves cutting quality and efficiency, and provides a new technical solution for precision machining of brittle materials.

[0253] With the above embodiments of the present invention as inspiration, and through the above description, relevant personnel can make various changes and modifications without departing from the technical scope of this invention. The technical scope of this invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A method for controlling thermal stress in laser cutting of brittle materials, characterized in that: The method comprises: S1, obtains real-time temperature distribution data of brittle materials during laser cutting. The surface temperature changes of the material are captured by infrared thermal imaging technology. Based on the influence of heat accumulation and steep temperature gradient, the temperature gradient value is calculated to obtain the initial thermal distribution characteristics. S2, extract the local thermal stress concentration area from the initial thermal distribution characteristics, use the finite element analysis method to simulate the thermal stress distribution under the thermal stress runaway state, combine the dynamic data of material properties, determine the potential location and range of microcrack propagation, and determine the thermal stress concentration coefficient; S3, based on the thermal stress concentration coefficient and temperature gradient value, construct a mathematical model of the relationship between thermal stress and material fracture, introduce dynamic parameters of material properties, and iteratively calculate the thermal stress runaway threshold through the lambda algorithm to obtain the thermal stress control boundary conditions; S4, adjust the laser parameter matching scheme according to the thermal stress control boundary conditions, obtain the laser power and pulse frequency combination that dynamically adapts to the material properties through the pre-established laser energy distribution database, and determine the optimized laser parameters; S5, through the optimized laser parameters and combined with auxiliary process optimization technology, introduces synchronous cooling airflow during the cutting process to control the steepness of the temperature gradient, and uses the airflow pressure regulation algorithm to dynamically adjust the airflow intensity to obtain a gentle temperature gradient distribution; S6, extracting the probability of thermal shock induction from the gentle temperature gradient distribution. If the probability of thermal shock induction exceeds a preset threshold value of thermal shock induction probability, the cutting speed adjustment parameter is adjusted to reduce the influence of heat accumulation, determine the degree of reduction in real-time fracture risk, and determine stable cutting conditions; S7, based on the stable cutting conditions and dynamic data of material properties, trains the parameter prediction model, uses a convolutional neural network to process the historical data of temperature distribution, thermal stress distribution and laser parameters, and outputs the optimal laser parameter combination that matches the characteristics of the brittle material; S8, after obtaining the optimal laser parameter combination, is applied in real time in the fully automatic laser cutting system. The microcrack diffusion state is monitored by sensors. If the microcrack diffusion area exceeds the preset microcrack diffusion area threshold, the auxiliary process optimization parameters are fine-tuned to obtain the final cutting quality control solution.

2. The method according to claim 1, characterized in that Step S1, obtaining real-time temperature distribution data of the brittle material during laser cutting, capturing the surface temperature change of the material using infrared thermal imaging technology, calculating the temperature gradient value based on the influence of heat accumulation and the steep temperature gradient, and obtaining the initial thermal distribution characteristics, includes: Step S11, collecting surface temperature data of the brittle material during laser cutting by infrared thermal imaging technology to obtain real-time temperature distribution; Step S12, using a two-dimensional convolution algorithm to process the real-time temperature distribution and extract the temperature change trend; Step S13, calculating the temperature gradient of the heat accumulation area based on the extracted temperature change trend to obtain a gradient value distribution; Step S14: using the principal component analysis method to obtain the characteristic vector of the initial thermal distribution based on the gradient value distribution, and determining the influence range of the thermal accumulation; Step S15, if the temperature gradient in the feature vector exceeds the temperature gradient threshold, the initial thermal distribution is smoothed by mean filtering to obtain an optimized thermal distribution; Step S16, extracting distribution features from the optimized thermal distribution to determine the thermal stress concentration area; Step S17: classify the thermal stress concentration areas using a support vector machine algorithm, use radial basis function as a kernel function, train a classification model, and determine the potential crack location.

3. The method according to claim 2, characterized in that The step S17, classifying the thermal stress concentration area by a support vector machine algorithm, using a radial basis function as a kernel function, training a classification model, and determining the potential crack location, includes: Step S171 , collecting temperature distribution of thermal stress concentration areas through thermal imaging data to obtain preliminary area division results; Step S172: grouping the preliminary region division results using a K-means clustering algorithm to determine the boundary range of the thermal stress concentration; Step S173, extracting spatial feature data from the boundary range to obtain a feature vector of thermal stress distribution; Step S174, classifying the feature vectors using a support vector machine algorithm, using a radial basis function as a kernel function, and training a classification model; Step S175: If the confidence level output by the classification model exceeds the confidence threshold, a high-risk area for potential cracks is determined; Step S176: For high-risk areas, a sliding window method is used to obtain the local temperature change trend and determine the crack propagation direction; Step S177: extract dynamic features from the crack propagation direction and determine the priority order of crack development.

4. The method according to claim 1, wherein Step S3, based on the thermal stress concentration coefficient and the temperature gradient value, constructs a mathematical model of the relationship between thermal stress and material fracture, introduces dynamic parameters of material properties, and iteratively calculates the thermal stress runaway threshold through the lambda algorithm to obtain the thermal stress control boundary conditions, including: Step S31, based on the thermal stress concentration and temperature gradient data, a mathematical model is constructed using the finite element method to calculate the thermal stress value distribution; Step S32: inputting the thermal stress value distribution result into the Laplace iterative algorithm for iterative calculation to determine the dynamic parameter change trend; Step S33, adjusting the thermal stress value using the material constitutive relationship according to the dynamic parameter change trend; Step S34: if the adjusted thermal stress value exceeds the preset thermal stress runaway threshold, recalculate the control boundary using a mathematical model; Step S35, by regulating the boundary to limit the thermal stress value, using the fracture mechanics criterion to determine the conditions for material fracture; Step S36, obtaining the concentrated distribution of thermal stress under the material fracture condition, and iteratively updating the thermal stress out-of-control threshold using the least squares method; Step S37: Using the updated thermal stress runaway threshold, determine the final control boundary condition.

5. The method according to claim 1, wherein The step S4, adjusting the laser parameter matching scheme based on the thermal stress control boundary conditions, obtaining the laser power and pulse frequency combination dynamically adapted to the material properties through the pre-established laser energy distribution database, and determining the optimized laser parameters, includes: Step S41, obtaining initial values ​​of laser parameters related to boundary conditions through a pre-established laser energy distribution database to obtain a preliminary matching solution; Step S42: In the preliminary matching scheme, material characteristic data is introduced, dynamic adaptation characteristics are extracted, and the corresponding laser power range is determined; Step S43, after determining the laser power range, using a binary search algorithm to adjust the pulse frequency to obtain a frequency combination that matches the laser power; Step S44: If the thermal stress control requirements change, re-match the laser parameters and update the matching solution based on the energy distribution data in the database; Step S45, obtaining laser parameters that match the boundary conditions according to the updated matching scheme, and determining an adjusted thermal stress control scheme; Step S46: In the thermal stress control scheme, linear regression is used to analyze the correlation between material properties and laser power to determine the optimization direction of production parameters; Step S47: According to the parameter optimization direction, the ratio of pulse frequency to laser power is adjusted to determine the final laser parameters.

6. The method according to claim 5, characterized in that The step S46, in the thermal stress control scheme, uses linear regression to analyze the correlation between material properties and laser power to determine the direction of production parameter optimization, including: Step S461: Acquire characteristic data related to thermal stress regulation through a pre-established material property database to determine an initial analysis range; Step S462: For the initial analysis range, linear regression is used to analyze the correlation between material characteristics and laser power to obtain characteristic change trends; Step S463: extracting a production parameter set that matches the control scheme from the energy distribution data based on the characteristic change trend, and determining the production parameter screening direction; Step S464: If the production parameter screening direction deviates from the expectation, the material characteristic data is updated through the database to obtain an adjusted characteristic set; Step S465: cluster analysis is performed on the adjusted feature set to prioritize production parameter adjustments and determine an optimized production parameter combination. Step S466, adjusting the boundary conditions of the control scheme by optimizing the production parameter combination to obtain a balanced scheme for thermal stress distribution; Step S467: According to the balancing solution, the final production parameter ratio is extracted from the matching solution to determine the execution parameters of the thermal stress control.

7. The method according to claim 1, characterized in that Step S6, extracting the thermal shock induction probability from the gentle temperature gradient distribution, and if the thermal shock induction probability exceeds a preset thermal shock induction probability threshold, adjusting the cutting speed adjustment parameter to reduce the impact of heat accumulation, determining the degree of reduction in real-time fracture risk, and determining stable cutting conditions, includes: Step S61, obtaining temperature gradient distribution data through a sensor, and calculating the probability of thermal shock induction using kernel density estimation to obtain a probability value; Step S62: If the probability value exceeds the thermal shock induction probability threshold, the relationship between thermal shock and heat accumulation is predicted by linear regression to determine the cutting speed adjustment range; Step S63, updating the cutting speed parameter according to the adjustment range, using a real-time monitoring tool to obtain the trend of heat accumulation changes, and determining the degree of heat accumulation reduction; Step S64, calculating the fracture risk change rate based on the relationship between the degree of heat accumulation reduction and the fracture risk, to obtain a risk reduction index; Step S65: If the risk reduction index reaches the stable threshold, the stable cutting state is confirmed by the ARIMA model to determine the validity of the current cutting speed parameter; Step S66, obtaining real-time fracture risk data, comparing the risk reduction index with the stability threshold, and determining whether further adjustment of the cutting speed parameters is required; Step S67, by iterating the above process in a loop, the temperature gradient data is updated using kernel density estimation to obtain the optimized stable cutting conditions.

8. The method according to claim 1, characterized in that The step S7, based on the stable cutting conditions and the dynamic data of the material properties, trains the parameter prediction model, uses a convolutional neural network to process the historical data of the temperature distribution, thermal stress distribution and laser parameters, and outputs the optimal laser parameter combination that matches the brittle material properties, including: Step S71, processing historical data through a convolutional neural network, inputting historical data as time series temperature and thermal stress data, using the TensorFlow framework for training, and obtaining initial laser parameter prediction results; Step S72, extracting temperature distribution data from the initial laser parameter prediction results, calculating the temperature distribution using a finite element analysis tool, and obtaining characteristic data that matches stable cutting conditions; Step S73, using ANSYS software to calculate the thermal stress distribution based on the characteristic data, and determine a response value consistent with the characteristics of the brittle material; Step S74: if the response value exceeds a preset response value threshold, adjusting the laser power and scanning speed according to the difference between the response value and the response value threshold to obtain a corrected laser parameter set; Step S75: recalculating the temperature distribution and thermal stress distribution using a finite element analysis tool for the corrected laser parameter set to determine whether the stable cutting requirements are met; Step S76, based on the verification results, the training model is optimized using the TensorFlow framework to obtain the laser parameter output that matches the optimal combination; Step S77: Use the optimized laser parameter output to process the real-time dynamic data and determine the final laser parameter combination solution.

9. The method according to claim 1, characterized in that In step S8, after obtaining the optimal laser parameter combination, the optimal laser parameter combination is applied in real time in the fully automatic laser cutting system. The microcrack diffusion state is monitored by a sensor. If the microcrack diffusion area exceeds a preset microcrack diffusion area threshold, the auxiliary process optimization parameters are fine-tuned to obtain the final cutting quality control solution, including: Step S81, after obtaining the laser parameter data, the optimal combination is calculated using the least square method to obtain an initial laser parameter set; Step S82, loading the initial laser parameter set into the fully automatic cutting system, and generating cutting path data through real-time application; Step S83, collecting cutting path data through a sensor, and using an image processing algorithm to extract the microcrack diffusion state from the collected data to obtain a diffusion area value; Step S84, if the diffusion area value exceeds the microcrack diffusion area threshold, the auxiliary process parameters are adjusted by analyzing the diffusion state, and an optimized process parameter group is obtained using an optimization algorithm; Step S85, updating the cutting system with the optimized process parameter set, and generating adjusted cutting data through real-time application; Step S86, monitoring the adjusted cutting data through sensors, using image recognition algorithms to determine the microcrack diffusion state from the monitoring results, and obtaining final cutting quality parameters; Step S87: Adjust the laser parameters according to the final cutting quality parameters, and obtain a stable control solution through iterative calculation.

10. A thermal stress control system for laser cutting of brittle materials, characterized in that: Thermal stress control is performed on laser cutting of brittle materials using the method according to any one of claims 1 to 9, the system comprising: The temperature distribution acquisition module is used to obtain real-time temperature distribution data of brittle materials during laser cutting. It uses infrared thermal imaging technology to capture the temperature changes on the material surface. Based on the influence of heat accumulation and steep temperature gradients, it calculates the temperature gradient value and obtains the initial thermal distribution characteristics. The stress concentration analysis module is used to extract local thermal stress concentration areas from the initial thermal distribution characteristics, simulate the thermal stress distribution under the thermal stress runaway state using the finite element analysis method, and determine the potential location and range of microcrack propagation and the thermal stress concentration coefficient by combining the dynamic data of material properties; The thermal stress modeling module is used to construct a mathematical model of the relationship between thermal stress and material fracture based on the thermal stress concentration coefficient and temperature gradient value. It introduces dynamic parameters of material properties and iteratively calculates the thermal stress runaway threshold through the lambda algorithm to obtain the thermal stress control boundary conditions. The laser parameter optimization module is used to control boundary conditions based on thermal stress, adjust the laser parameter matching scheme, obtain the laser power and pulse frequency combination that dynamically adapts to the material properties through the pre-established laser energy distribution database, and determine the optimized laser parameters; The temperature gradient control module is used to control the steepness of the temperature gradient by introducing synchronous cooling airflow during the cutting process through optimized laser parameters and combined with auxiliary process optimization technology. The airflow intensity is dynamically adjusted using an airflow pressure regulation algorithm to obtain a smooth temperature gradient distribution. The cutting condition determination module is used to extract the probability of thermal shock induction from the gentle temperature gradient distribution. If the probability of thermal shock induction exceeds the preset thermal shock induction probability threshold, the cutting speed adjustment parameter is adjusted to reduce the impact of heat accumulation, determine the degree of real-time fracture risk reduction, and determine stable cutting conditions; The parameter prediction training module is used to train the parameter prediction model based on stable cutting conditions and dynamic data of material properties. It uses a convolutional neural network to process historical data of temperature distribution, thermal stress distribution and laser parameters, and outputs the optimal laser parameter combination that matches the characteristics of brittle materials. The quality control execution module is used to obtain the optimal laser parameter combination and apply it in real time in the fully automatic laser cutting system. It monitors the microcrack diffusion status through sensors. If the microcrack diffusion area exceeds the preset microcrack diffusion area threshold, the auxiliary process optimization parameters are fine-tuned to obtain the final cutting quality control plan.

Images