Intelligent control method and system for aluminum profile production and machining process

Abstract

The invention discloses an intelligent control method and system for an aluminum profile production and processing process, relates to the field of intelligent control of the aluminum profile production and processing process, is used for intelligently controlling the aluminum profile production and processing process and effectively reducing the internal stress risk, and comprises the following steps: acquiring various production process parameters related to the internal stress accumulation of a profile; according to the various production process parameters, risk measurement of internal stress accumulation of the profile is obtained; generating a comprehensive evaluation result based on the risk measurement and the energy consumption information of the production process, the comprehensive evaluation result being used for evaluating the relationship between the quantitative production efficiency and the risk measurement; and according to the comprehensive evaluation result, the production process parameters are adjusted, so that the balance between the production efficiency and the risk measurement is realized on the premise of ensuring that the surface quality and the instant performance of the profile meet the requirements.

Description

Technical Field

[0001] This invention relates to the field of intelligent control of aluminum profile production and processing, and in particular to an intelligent control method and system for aluminum profile production and processing. Background Technology

[0002] In the intelligent extrusion production of aluminum profiles, although intelligent control systems are widely used to optimize production processes and improve efficiency, the accumulation of environmental conditions and equipment performance degradation during long-term operation may induce invisible, high-level residual stresses within the profile during compensatory adjustments. These residual stresses are difficult to detect through conventional testing, yet they significantly affect the subsequent processing performance and long-term service reliability of the profiles, such as causing processing deformation, reducing fatigue life, and increasing the risk of stress corrosion cracking. Existing intelligent control systems lack the ability to directly monitor and predict residual stresses, thus failing to identify and proactively suppress the generation of this potential defect. Summary of the Invention

[0003] This application discloses an intelligent control method and system for the production and processing of aluminum profiles, which aims to solve the technical problem that in the existing intelligent extrusion production process of aluminum profiles, the lack of direct monitoring and prediction capabilities for residual stress inside the profiles leads to the system's inability to identify and actively suppress the generation of potential defects, thereby affecting the subsequent processing performance and long-term service reliability of the profiles.

[0004] In a first aspect, this application discloses an intelligent control method for the production and processing of aluminum profiles, comprising the following steps: Acquire various production process parameters related to the accumulation of internal stress in the profile; these parameters include the surface temperature distribution of the profile at the outlet of the cooling zone, the difference between the actual operating power and the theoretical required power of the cooling fan, the complexity of the cross-sectional geometry of the current profile, and the fluctuation of the extrusion speed under stable operating conditions. Based on various production process parameters, a risk measure for the accumulation of internal stress in the profile is obtained; Based on risk measurement and energy consumption information of the production process, a comprehensive evaluation result is generated. The comprehensive evaluation result is used to assess the relationship between quantitative production efficiency and risk measurement. Based on the comprehensive evaluation results, the production process parameters were adjusted to achieve a balance between production efficiency and risk measurement while ensuring that the surface quality and immediate performance of the profiles meet the requirements.

[0005] Optionally, a risk measure for internal stress accumulation in the profile can be obtained based on various production process parameters, including: Based on various production process parameters and their corresponding mapping relationships, determine the risk value corresponding to each production process parameter; Based on the weights corresponding to each production process parameter, the risk values ​​corresponding to each production process parameter are weighted and summed to obtain a risk measure of internal stress accumulation in the profile.

[0006] Optionally, based on the weights corresponding to each production process parameter, a weighted summation of the risk values ​​corresponding to each production process parameter is performed to obtain a risk measure for the accumulation of internal stress in the profile, including: Based on the weights corresponding to each production process parameter, the risk values ​​corresponding to each production process parameter are weighted and summed to obtain the initial risk measure of internal stress accumulation in the profile. The cross-section of the profile is divided into multiple micro-regions, and the geometric thermal sensitivity index of each micro-region is calculated. The geometric thermal sensitivity index is used to reflect the tendency of the micro-region to induce deep stress when the cooling is uneven. For any micro-region, when the deviation ratio between the actual temperature decrease rate or temperature fluctuation frequency of the micro-region and the expected dynamic cooling trajectory is greater than a deviation threshold, and the deviation exhibits non-periodic characteristics, a transient heat exchange anomaly is identified in the micro-region; and Based on the geometric thermal sensitivity index of regions with transient heat exchange anomalies, the initial risk metric is corrected to obtain the risk metric.

[0007] Optionally, the method also includes: Local temperature gradient analysis and high-frequency fluctuation feature extraction are performed on surface temperature distribution information. Trend analysis and abnormal duration assessment are conducted on the difference between the actual operating power and theoretical required power of the cooling fan. The complexity of the cross-sectional geometry of the current production profile is quantified by wall thickness difference and irregularity index. Spectral analysis is performed on the fluctuation information of extrusion speed under stable operating conditions to obtain the parameter characteristics of the production process parameters. The irregularity index is the area of ​​the cross-section and the minimum circumscribed rectangle or circle. Multi-dimensional correlation analysis of parameter characteristics is performed to identify deep stress accumulation modes caused by the synergistic effect of multiple parameters; Based on the deep stress accumulation model, the initial weights corresponding to each production process parameter are adjusted to obtain the weights corresponding to each production process parameter.

[0008] Optionally, based on the comprehensive evaluation results, production process parameters may be adjusted, including: Obtain the alloy grade, cross-sectional geometry, mechanical performance requirements of the target product, and expected production cycle of the profile; Based on the alloy grade, cross-sectional geometry, mechanical performance requirements of the target product, and expected production cycle, a set of nonlinear relational functions describing the effects of cooling intensity and extrusion speed on the internal stress accumulation of the profile and the energy consumption of the production process is extracted. By substituting risk metrics and energy consumption information into a set of nonlinear relational functions, the combined impact of the current combination of production process parameters on the internal stress accumulation of profiles and the energy consumption of the production process is evaluated, and the evaluation results are obtained. Based on the evaluation results, the cooling intensity and extrusion speed are adjusted until a combination of production process parameters is found that achieves a global balance between the risk measure of internal stress accumulation in the profile and the energy consumption of the production process.

[0009] Optionally, the method also includes: Record the current combination of process parameters, risk measurement of internal stress accumulation in the profile, and energy consumption information of the production process; Compare the current combination of process parameters with the local optimal solutions on the historical optimization path to determine whether the optimization process has fallen into a local state; When the optimization process gets stuck in a local state, adjust the set of nonlinear relational functions; Based on the adjusted set of nonlinear relational functions, the evaluation and adjustment are carried out again until a combination of production process parameters that can achieve a global balance between the risk measure of internal stress accumulation in the profile and the energy consumption of the production process is found.

[0010] Optionally, when the optimization process gets stuck in a local state, the set of nonlinear relational functions is adjusted, including: When the optimization process gets stuck in a local state, increase the fine-tuning step size; Based on the adjusted fine-tuning step size, the key parameters in the set of nonlinear relational functions are fine-tuned.

[0011] Optionally, when the optimization process gets stuck in a local state, the set of nonlinear relational functions is adjusted, including: Obtain the current combination of process parameters for the optimization iteration; Based on the current combination of process parameters, risk metrics, and energy consumption information in the optimization iteration, calculate the curvature change and gradient information of the optimization trajectory; Based on the curvature change and gradient information of the optimization trajectory, determine whether the optimization process has entered the saddle point region or the flat region. When the optimization process enters the saddle point region, random disturbances are introduced to change the current combination of process parameters and escape the saddle point region. When the optimization process is determined to have entered a flat region, the exploratory step size is increased to expand the parameter search range; Adjust the set of nonlinear relational functions based on the changed combination of current process parameters or the parameter combination after expanding the parameter search range.

[0012] Optionally, when the optimization process enters the saddle point region, a random disturbance is introduced to change the current combination of process parameters and escape the saddle point region, including: Obtain the alloy grade, cross-sectional geometry, and mechanical property requirements of the target product for the profile; Based on the alloy grade, cross-sectional geometry, and mechanical property requirements of the target product, determine the instantaneous safety fluctuation boundaries of cooling intensity and extrusion speed; Within the instantaneous safety fluctuation boundary, a random disturbance is generated; The random disturbance is superimposed on the current cooling intensity and extrusion speed parameters to form an exploratory parameter combination; The effects of preliminary experimental parameter combinations on the surface temperature distribution at the outlet of the cooling zone of the profile, the instantaneous tensile strength and yield strength of the profile; When the prediction results show that the cooling intensity and extrusion speed are within the instantaneous safety fluctuation boundary, an exploratory combination of parameters is applied to the production line to change the current combination of process parameters and escape the saddle point region.

[0013] Secondly, this application also discloses an intelligent control system for the production and processing of aluminum profiles, the system comprising: The parameter acquisition module is used to acquire various production process parameters related to the internal stress accumulation of the profile. These parameters include the surface temperature distribution of the profile at the cooling zone outlet, the difference between the actual operating power and the theoretical required power of the cooling fan, the complexity of the cross-sectional geometry of the current profile, and the fluctuation of the extrusion speed under stable operating conditions. The risk measurement module is used to measure the risk of internal stress accumulation in profiles based on various production process parameters. The comprehensive assessment module is used to generate comprehensive assessment results based on risk measurement and energy consumption information of the production process. The comprehensive assessment results are used to evaluate the relationship between quantitative production efficiency and risk measurement. The parameter adjustment module is used to adjust the production process parameters based on the comprehensive evaluation results, so as to achieve a balance between production efficiency and risk measurement while ensuring that the surface quality and immediate performance of the profiles meet the requirements.

[0014] Beneficial effects This application discloses an intelligent control method for aluminum profile manufacturing. It acquires various production process parameters related to internal stress accumulation in the profile, including surface temperature distribution at the cooling zone outlet, the difference between the actual operating power and theoretical power requirement of the cooling fan, the complexity of the current profile cross-sectional geometry, and fluctuations in extrusion speed under stable operating conditions. Based on these parameters, a risk measure for internal stress accumulation is obtained. Furthermore, a comprehensive evaluation result is generated based on the risk measure and energy consumption information of the production process to assess the relationship between production efficiency and risk measure. Finally, based on the comprehensive evaluation result, production process parameters are adjusted to achieve a balance between production efficiency and risk measure while ensuring that the profile surface quality and immediate performance meet requirements.

[0015] This method effectively solves the problem of existing intelligent control systems lacking the ability to directly monitor and predict residual stress. By introducing multi-dimensional production process parameters and combining them with risk metrics and energy consumption information for comprehensive evaluation, this application can assess the level of residual stress inside the profile in real time or near real time and proactively adjust the extrusion process parameters accordingly. This not only ensures that the surface quality and immediate mechanical properties of the profile meet requirements, but more importantly, it can effectively control and reduce the high residual stress hidden inside the profile, avoiding problems such as deformation, reduced fatigue life, and stress corrosion cracking caused by it in subsequent processing and long-term service. Compared with existing technologies, this application does not require the addition of expensive and complex online residual stress detection equipment. Instead, it utilizes existing or a small number of easily deployable sensor data, combined with advanced data analysis and modeling methods, to optimize the performance of the product throughout its entire life cycle, demonstrating significant technological progress and practical value. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of an intelligent control method for aluminum profile production and processing provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of another intelligent control method for aluminum profile production and processing provided in this embodiment of the invention; Figure 3 This is a schematic diagram of the structure of an intelligent control system for aluminum profile production and processing provided in an embodiment of the present invention. Detailed Implementation

[0017] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0018] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0019] First, let's introduce the terminology used in this application.

[0020] "Risk metric for internal stress accumulation in profiles" refers to an indicator that assesses the likelihood and severity of residual stress generation and accumulation in aluminum profiles during the production process by quantifying and analyzing various production process parameters. The higher the value of this metric, the higher the level of internal residual stress in the profile or the greater the risk of generating high residual stress.

[0021] "Energy consumption information in the production process" refers to the total amount of various energy sources, such as electrical energy and thermal energy, consumed in the aluminum profile production and processing process, including extrusion, cooling, and other stages, or the energy consumption per unit output.

[0022] The "Comprehensive Assessment Result" is an output that comprehensively evaluates the current production status by combining risk measures of internal stress accumulation in the profiles with energy consumption information from the production process. This result aims to quantify the relationship between production efficiency and risk measures, providing a basis for subsequent adjustments to process parameters.

[0023] "Production process parameters" refer to various process variables that can be adjusted and controlled during the production of aluminum profiles, such as extrusion speed, cooling intensity (including cooling air volume, water flow, etc.), and mold temperature.

[0024] The core of the intelligent control method for aluminum profile production and processing proposed in this application lies in acquiring, analyzing and evaluating various production process parameters related to the accumulation of internal stress in the profile, and then intelligently adjusting the production process parameters to achieve a balance between production efficiency and risk measurement of internal stress accumulation in the profile.

[0025] The following specific embodiments will provide a detailed introduction and explanation of an intelligent control method for aluminum profile production and processing provided in this application.

[0026] Reference Figure 1 This invention provides an intelligent control method for the production and processing of aluminum profiles, comprising the following steps: S1 obtains various production process parameters related to the accumulation of internal stress in the profile.

[0027] Among them, various production process parameters include information on the surface temperature distribution of the profile at the outlet of the cooling zone, information on the difference between the actual operating power and the theoretical required power of the cooling fan, information on the complexity of the cross-sectional geometry of the currently produced profile, and information on the fluctuation of the extrusion speed under stable operating conditions.

[0028] Specifically, infrared thermal imagers or distributed temperature sensor arrays can be installed at the cooling zone outlet to collect real-time information on the surface temperature distribution of the profiles. These sensors can acquire temperature field data of the profile surface in a non-contact manner and transmit it to the data processing unit. Furthermore, the difference between the actual operating power and the theoretical required power of the cooling fan can be obtained by monitoring the power consumption and airflow output of the cooling fan and comparing it with a preset theoretical model. Information on the complexity of the geometry of the current production profile cross-section can be analyzed using CAD model data or image recognition technology to quantify its geometric characteristics such as wall thickness variation and irregularity. Fluctuations in the extrusion speed under stable operating conditions can be monitored in real-time using the extruder's main motor encoder or speed sensor, and the collected speed data can be statistically analyzed, such as calculating the standard deviation or fluctuation frequency.

[0029] S2. Based on various production process parameters, obtain a risk measure for the accumulation of internal stress in the profile.

[0030] The value assessment results are used to measure the importance of operational tasks or external power supply requests.

[0031] Specifically, the risk value corresponding to each production process parameter can be determined based on various production process parameters and their corresponding mapping relationships; based on the weights corresponding to each production process parameter, the risk values ​​corresponding to each production process parameter are weighted and summed to obtain a risk measure of internal stress accumulation in the profile.

[0032] The mapping relationship can be understood as converting the raw data or characteristic values ​​of each production process parameter into a standardized risk score or risk level function or rule set. For example, for surface temperature distribution information at the cooling zone outlet, when the temperature distribution non-uniformity exceeds a certain threshold, the corresponding risk value will increase with the increase of non-uniformity. The risk value aims to quantify the potential contribution or influence of a single production process parameter on the internal stress accumulation of the profile.

[0033] Furthermore, weights refer to the importance coefficients assigned to each production process parameter, reflecting the relative importance of that parameter in influencing the risk of internal stress accumulation in the profile. These weights can be determined based on expert experience, historical data analysis, machine learning model training results, or sensitivity analysis. For example, if historical data shows that the difference between the actual operating power and the theoretical required power of the cooling fan has the most significant impact on internal stress accumulation, then the weight of that parameter will be set relatively high. Weighted summation involves multiplying the risk value corresponding to each production process parameter by its corresponding weight, and then summing all the products to obtain a comprehensive risk measure.

[0034] S3. Generate a comprehensive assessment result based on risk measurement and energy consumption information of the production process.

[0035] The comprehensive evaluation results are used to assess the relationship between quantitative production efficiency and risk measurement.

[0036] Specifically, a multi-objective optimization function can be constructed, where one objective minimizes the risk metric, and the other minimizes energy consumption or maximizes production efficiency. By substituting the risk metric and energy consumption information into this function, a comprehensive score can be obtained, which is the overall evaluation result. This evaluation result can intuitively reflect the level of internal stress accumulation risk and the corresponding energy consumption cost undertaken to achieve a certain production efficiency under the current production state.

[0037] S4. Based on the comprehensive evaluation results, adjust the production process parameters to achieve a balance between production efficiency and risk measurement while ensuring that the surface quality and immediate performance of the profiles meet the requirements.

[0038] For example, when the comprehensive assessment results indicate that the risk metric is too high, even if production efficiency is high, the system may choose to reduce the extrusion speed or adjust the cooling intensity to reduce the risk of internal stress accumulation in the profile. Conversely, when the risk metric is low and energy consumption is high, the system may appropriately increase the extrusion speed or optimize the cooling strategy, while ensuring that the risk is controllable, to improve production efficiency or reduce energy consumption. This adjustment process can be achieved through preset control strategies, fuzzy logic controllers, or reinforcement learning-based agents, thereby dynamically optimizing production process parameters.

[0039] The intelligent control method for aluminum profile manufacturing proposed in this application works by constructing a closed-loop intelligent control system to solve the problem of balancing production efficiency, surface quality, immediate performance, and internal residual stress control in traditional aluminum profile production. Existing intelligent control systems often focus on achieving surface quality and immediate performance targets, neglecting the potential for high levels of residual stress induced within the profile when compensating for external changes. While these residual stresses are not immediately apparent, they pose a potential threat to subsequent processing and long-term service reliability of the profile.

[0040] This application introduces the core concept of "risk measurement of internal stress accumulation in profiles" and combines it with energy consumption information from the production process to generate a comprehensive assessment result, thereby achieving a comprehensive quantitative evaluation of the production process. Specifically, firstly, through the parameter acquisition module, the system can acquire, in real-time or near real-time, information on the surface temperature distribution of the profile at the cooling zone outlet, the difference between the actual operating power and the theoretical required power of the cooling fan, the complexity of the current profile cross-sectional geometry, and the fluctuation of the extrusion speed under stable operating conditions. These parameters are carefully selected because they have a direct or indirect physical correlation with the formation and accumulation of internal stress in the profile. For example, uneven surface temperature distribution can lead to internal temperature gradients, thereby generating thermal stress; differences in cooling fan power may reflect abnormal cooling efficiency, affecting cooling uniformity; complex cross-sectional shapes are inherently prone to stress concentration during cooling; and fluctuations in extrusion speed can affect the stability of heat exchange.

[0041] Next, the risk measurement module utilizes these multi-dimensional parameters and, through a pre-defined model or algorithm, calculates a risk measure for the accumulation of internal stress in the profile. This measure is a quantitative assessment of the potential residual stress level, making the invisible internal stress risk perceptible and manageable. Subsequently, the comprehensive assessment module combines this risk measure with energy consumption information from the production process to generate a comprehensive assessment result. This result not only considers production efficiency and energy consumption, but more importantly, it quantifies the relationship between production efficiency and internal stress risk, providing decision-makers with a basis for balancing different objectives.

[0042] Finally, the parameter adjustment module intelligently adjusts production process parameters, such as extrusion speed and cooling intensity, based on the comprehensive evaluation results. This adjustment does not simply pursue a single objective, but rather seeks the optimal balance between production efficiency and internal stress risk while ensuring that the profile surface quality and immediate performance meet requirements. For example, when the risk level is too high, the system may sacrifice some production efficiency to reduce stress risk; conversely, when the risk level is within an acceptable range, the system may optimize parameters to improve efficiency or reduce energy consumption. The entire process forms a dynamic feedback loop, enabling the system to adaptively respond to various changes in the production process, thereby effectively controlling and reducing residual stress within the profile and optimizing the product's performance throughout its entire lifecycle.

[0043] The intelligent control method for aluminum profile manufacturing proposed in this application, compared to existing technologies, has a core innovation in introducing a quantitative assessment and balance control mechanism for the risk of internal stress accumulation in the profile. Traditional intelligent control systems primarily focus on the surface quality and immediate mechanical properties of the profile, adjusting process parameters to ensure these indicators meet standards. However, this strategy often overlooks the problem of high-level residual stresses that may be induced within the profile when compensating for changes in the external environment or equipment. These residual stresses have a significant negative impact on the subsequent processing performance and long-term service reliability of the profile, but due to their hidden nature, they are difficult to detect and control in existing systems.

[0044] This application constructs a mechanism capable of real-time or near-real-time assessment of internal stress risk by acquiring various production process parameters related to internal stress accumulation, including surface temperature distribution of the profile at the cooling zone outlet, the difference between the actual operating power and theoretical required power of the cooling fan, the complexity of the current profile cross-sectional geometry, and fluctuations in extrusion speed under stable operating conditions. The selection of these parameters has profound physical significance, directly or indirectly reflecting the stress state that the profile may experience during its thermomechanical history. For example, uneven surface temperature distribution is a direct cause of internal thermal stress; differences in cooling fan power may indicate fluctuations in cooling efficiency, thus affecting cooling uniformity; complex cross-sectional shapes themselves are prone to stress concentration during cooling; and fluctuations in extrusion speed affect the material's deformation behavior and the stability of the heat exchange process.

[0045] By comprehensively analyzing these parameters, this application can obtain a risk measure of internal stress accumulation in profiles, thus transforming the traditionally difficult-to-measure internal stress risk into a quantifiable indicator. Based on this, this application further combines this risk measure with energy consumption information from the production process to generate a comprehensive evaluation result, used to assess the relationship between quantitative production efficiency and risk measurement. This innovative comprehensive evaluation mechanism enables the system to proactively consider and control internal stress risk while pursuing production efficiency and reducing energy consumption, avoiding the drawbacks of traditional systems that may sacrifice internal quality for surface quality and immediate performance targets.

[0046] Ultimately, based on the comprehensive evaluation results, this application intelligently adjusts the production process parameters to achieve a balance between production efficiency and risk measurement while ensuring that the surface quality and immediate performance of the profiles meet the requirements. This balance control strategy is not available in existing technologies, enabling the aluminum profile production process to move from "standard-compliant production" to "high-quality, high-reliability production," significantly improving the product's lifecycle performance. Compared to existing technologies, the advantage of this application lies in its ability to effectively manage and control internal stress risks without adding expensive and complex online residual stress detection equipment. It utilizes existing or minimally added easily deployable sensor data and employs advanced data analysis and modeling methods to provide a more comprehensive and intelligent control strategy for the intelligent manufacturing of aluminum profiles.

[0047] like Figure 2 As shown, in order to measure the risk of internal stress accumulation in the profile, this application may further include the following steps: S101. Based on the weights corresponding to each production process parameter, the risk values ​​corresponding to each production process parameter are weighted and summed to obtain the initial risk measure of internal stress accumulation in the profile.

[0048] The initial risk measure is determined by using the above-mentioned various production process parameters, through a preset mapping relationship to determine the risk value of each parameter, and by combining the weight of each parameter to perform a preliminary calculation, reflecting the stress accumulation risk at the macro level.

[0049] S102. Divide the cross-section of the profile into multiple micro-regions and calculate the geometric thermal sensitivity index of each micro-region.

[0050] Among them, the geometric thermal sensitivity index is used to reflect the tendency of micro-regions to induce deep stress when cooling is uneven.

[0051] Specifically, the cross-section of a profile can be discretized into a series of small, interconnected units, i.e., micro-regions. The geometric thermal sensitivity index can be calculated based on the geometric characteristics of each micro-region (such as wall thickness, curvature, contact area with the cooling medium, etc.) and its location within the entire cross-section. For example, thin-walled regions or regions with abrupt geometric changes typically have higher geometric thermal sensitivity indices because these regions are more prone to generating large temperature gradients during cooling, thereby inducing deep stresses.

[0052] S103. For any micro-region, when the deviation ratio between the actual temperature drop rate or temperature fluctuation frequency of the micro-region and the expected dynamic cooling trajectory is greater than the deviation threshold, and the deviation exhibits non-periodic characteristics, a transient heat exchange anomaly is identified in the micro-region.

[0053] The "expected dynamic cooling trajectory" refers to the temperature change path that this micro-region should follow under ideal or normal production conditions. When the actual temperature drop rate or temperature fluctuation frequency deviates significantly and non-periodically from this expected trajectory, it indicates that the region may have experienced sudden or unstable cooling conditions, such as sudden changes in local wind speed or uneven coolant spraying. These abnormal heat exchange events can easily induce transient thermal stress in local areas. The deviation threshold can be preset based on empirical data or simulation models to define what degree of deviation is considered abnormal.

[0054] S104. Based on the geometric thermal sensitivity index of regions with transient heat exchange anomalies, the initial risk measure is corrected to obtain the risk measure.

[0055] The operational cost index reflects the immediate impact of a specific operation on battery performance, while the battery asset depreciation factor transforms this immediate impact into a more economically meaningful long-term loss. This multiplicative operation quantifies the economic loss caused by a specific operation to the overall asset value of the battery.

[0056] Specifically, when transient heat exchange anomalies are identified in one or more micro-regions, the geometric thermal sensitivity indices of these regions are used to quantify their contribution to the overall stress accumulation risk. The correction process can involve weighting and superimposing the geometric thermal sensitivity indices of these anomalous regions with the initial risk measure, or adjusting them using a nonlinear function to increase the weight of these high-risk regions in the final risk measure. This allows the final risk measure to more accurately reflect the true risk of deep stress accumulation within the profile.

[0057] The solution presented in this application can more accurately assess the risk of internal stress accumulation in profiles because it incorporates a refined analysis of the profile cross-section at the microscopic level, building upon macroscopic risk assessment. First, by dividing the profile cross-section into multiple micro-regions and calculating their geometric thermal sensitivity indices, the system can identify localized weak areas more prone to deep stress under uneven cooling. This overcomes the limitation of traditional weighted summation methods of macroscopic parameters, which cannot distinguish differences in thermal response across different cross-sectional regions. Second, by real-time monitoring of the deviation between the actual temperature drop rate or temperature fluctuation frequency of the micro-regions and the expected dynamic cooling trajectory, and by identifying non-periodic transient heat exchange anomalies, this solution can promptly detect and locate thermal events that may lead to sudden localized stress concentrations. Since these transient anomalies are often key triggers for deep stress formation, their identification and quantification are crucial for accurate risk assessment. Finally, by incorporating the geometric thermal sensitivity index of these regions with transient heat exchange anomalies into the correction of the initial risk measurement, this scheme can effectively integrate local and transient stress accumulation risks into the overall risk assessment. This makes the final risk measurement not only reflect macro trends but also include micro details, significantly improving the accuracy and reliability of the risk assessment.

[0058] Through the above technical solution, this application provides a more refined and accurate method for measuring the risk of internal stress accumulation in profiles. Compared to relying solely on the weighted summation of macroscopic parameters, this solution, by introducing geometric thermal sensitivity analysis of micro-regions and identification of transient heat exchange anomalies, can effectively capture the risk of deep stress accumulation in localized areas within the profile caused by complex cooling conditions or sudden events. This makes the risk measurement no longer a simple average value, but rather reflects the potential problems in specific high-risk areas, thus providing a more targeted and effective basis for subsequent adjustments to production process parameters. Therefore, it can significantly reduce the risk of profile cracking, deformation, or performance degradation caused by internal stress accumulation, while optimizing production efficiency while ensuring product quality.

[0059] In some preferred embodiments, a specific example is given below. Suppose that when producing an aluminum profile with a complex cross-section (e.g., containing thin-walled and thick-walled regions), the initial risk metric is obtained by weighting and summing information on surface temperature distribution, the difference between the actual operating power and theoretical power requirement of the cooling fan, the complexity of the geometry of the current profile cross-section, and the fluctuation of the extrusion speed under stable operating conditions, and its value is 0.6 (range 0-1).

[0060] Specifically, the system first divides the cross-section of the profile into hundreds of micro-regions. For each micro-region, its geometric thermal sensitivity index is calculated based on its geometric characteristics, such as wall thickness and distance from the cooling air duct. For example, at a thin-walled joint of the profile, the geometric thermal sensitivity index is calculated to be 0.8, while in a thick-walled region, the index is 0.3.

[0061] During the production process, the temperature changes in these micro-regions are monitored in real time using infrared thermal imagers and local temperature sensors. At a certain moment, the system detects that the actual temperature drop rate of the aforementioned thin-walled connection area suddenly accelerates, and its deviation from the expected dynamic cooling trajectory exceeds a preset deviation threshold (e.g., 20%). At the same time, this deviation exhibits non-periodic characteristics, which is identified as a transient heat exchange anomaly.

[0062] Based on this, the system assigns a geometric thermal sensitivity index of 0.8 to the thin-walled connection region exhibiting transient heat exchange anomalies, and then modifies the initial risk measure of 0.6 based on the degree of anomaly. The correction algorithm can be a non-linear function, for example: Final risk measure = Initial risk measure + (Geometric thermal sensitivity index of the anomaly region) Anomaly severity factor). Assuming the anomaly severity factor is 0.3, then the final risk measure = 0.6 + (0.8) 0.3) = 0.6 + 0.24 = 0.84.

[0063] Through this correction, even though the initial risk metric of macroscopic parameters is not high, the final risk metric is significantly increased to 0.84 due to abnormal transient heat exchange and high geometric thermal sensitivity in localized areas. This prompts the control system to immediately adjust production process parameters such as cooling intensity and extrusion speed, for example, by reducing localized cooling airflow or slowing down the extrusion speed, to alleviate transient thermal stress in the thin-walled area. This effectively avoids the risk of profile cracking or deformation caused by localized stress concentration, ensuring product quality.

[0064] In some embodiments described above, the risk measurement of internal stress accumulation in the profile is obtained by weighted summation of risk values ​​corresponding to various production process parameters. However, in actual production, the influence of different production parameters on internal stress accumulation in the profile is not isolated, but may have complex synergistic effects, and their importance (i.e., weight) may also dynamically change with factors such as production conditions and profile characteristics. If fixed or empirical weights are used, it may be impossible to accurately identify the deep stress accumulation pattern caused by the synergistic effect of multiple parameters, thereby affecting the accuracy of risk measurement and the effectiveness of subsequent process adjustments.

[0065] In response, this application further proposes optimizations to the above method, which also include: S201. Perform local temperature gradient analysis and high-frequency fluctuation feature extraction on surface temperature distribution information; perform trend analysis and abnormal duration assessment on the difference between actual operating power and theoretical required power of cooling fan; quantify wall thickness difference and irregular shape index on the complexity of the cross-sectional geometry of the current production profile; and perform spectrum analysis on the fluctuation information of extrusion speed under stable operating conditions to obtain the parameter characteristics of production process parameters; the irregular shape index is the area of ​​the cross-section and the smallest circumscribed rectangle or circle.

[0066] Specifically, local temperature gradient analysis of surface temperature distribution information aims to identify abrupt changes in profile surface temperature in localized areas, which are often the direct cause of thermal stress concentration. High-frequency fluctuation feature extraction is used to capture rapid, transient changes in the temperature field, which may indicate instability in the cooling process. For the difference between the actual operating power and the theoretical required power of the cooling fan, trend analysis can reveal the long-term stability of the cooling system, while anomaly duration assessment focuses on the duration of abnormal cooling conditions, as prolonged anomalies have a more significant impact on profile quality.

[0067] The analysis quantifies the complexity of the cross-sectional geometry of the currently produced profiles by measuring wall thickness variation and irregularity index. Wall thickness variation refers to the difference in thickness between different regions of the profile cross-section; a large variation leads to uneven cooling rates. The irregularity index can be understood as the degree to which the profile cross-sectional shape deviates from a standard circle or rectangle, specifically defined as the ratio of the cross-section area to the area of ​​the smallest circumscribed rectangle or circle. Its purpose is to quantify the impact of cross-sectional shape on cooling uniformity and stress distribution. In practical applications, spectral analysis is performed on the fluctuation information of the extrusion speed under stable operating conditions to identify periodic or non-periodic disturbances during the extrusion process. These disturbances may induce stress by affecting material deformation and heat exchange processes. Through the above analysis, the parametric characteristics of the production process parameters can be obtained. These characteristics are a deeper and more physically meaningful abstraction of the original parameters.

[0068] S202. Perform multi-dimensional correlation analysis on parameter characteristics to identify deep stress accumulation modes caused by the synergistic effect of multiple parameters.

[0069] Furthermore, multi-dimensional correlation analysis of parameter characteristics aims to identify deep stress accumulation patterns caused by the synergistic effects of multiple parameters. For example, excessively large local temperature gradients, prolonged periods of abnormal cooling fan power, and high profile irregularity indices may collectively lead to deep stress accumulation in a specific area. This correlation analysis can be achieved using machine learning algorithms (such as decision trees, neural networks, or support vector machines) or statistical methods (such as principal component analysis and factor analysis) to reveal the nonlinear relationships and potential coupling effects between parameters.

[0070] S203. Based on the deep stress accumulation model, adjust the initial weights corresponding to each production process parameter to obtain the weights corresponding to each production process parameter.

[0071] When a specific stress accumulation pattern is identified, the weights of production process parameters that are strongly correlated with the pattern will be dynamically increased, while the weights of parameters with lower correlation may be appropriately reduced, so that the risk measurement model can more sensitively and accurately reflect the real risks under the current production conditions.

[0072] This application's solution, through in-depth feature extraction and multi-dimensional correlation analysis of various production process parameters, enables a more detailed and comprehensive understanding of the influence mechanism of each parameter on the internal stress accumulation of profiles. Traditional weighted summation methods, based solely on experience or preset static weights, struggle to capture the complex synergistic effects between parameters. This solution, however, identifies deep stress accumulation patterns caused by the synergistic effects of multiple parameters, revealing the underlying stress formation laws. It is precisely this identification of deep stress accumulation patterns that allows the weights of each production process parameter to be dynamically adjusted according to actual production conditions and stress patterns. This ensures that the risk measurement model more accurately reflects the true risk of current internal stress accumulation in the profiles, avoiding misjudgments or omissions due to inappropriate weighting.

[0073] In some preferred embodiments, a specific example is given below. Suppose that when producing an aluminum profile with a complex cross-section, the system first obtains information on the surface temperature distribution, the difference between the actual operating power and the theoretical required power of the cooling fan, the complexity of the geometry of the current profile cross-section, and the fluctuation of the extrusion speed under stable operating conditions.

[0074] Specifically, analysis of the surface temperature distribution revealed a persistent localized high-temperature gradient in the profile flange area, accompanied by high-frequency temperature fluctuations. Simultaneously, a persistent negative difference existed between the actual operating power of the cooling fan and the theoretically required power, indicating insufficient cooling intensity. Furthermore, the profile exhibited a high irregularity index and significant wall thickness variations. Spectral analysis of the extrusion speed fluctuations revealed periodic fluctuations at specific frequencies.

[0075] Through multi-dimensional correlation analysis, the system identified a deep stress accumulation mode: when "high irregularity index + local high temperature gradient + insufficient cooling + periodic extrusion speed fluctuation" work together, deep tensile stress is easily induced at the connection between thin and thick walls of the profile.

[0076] Based on the identified deep stress accumulation pattern, the system dynamically adjusts the weights of various parameters. For example, the weights of "surface temperature distribution information" and "difference between actual operating power and theoretical required power of the cooling fan" are significantly increased, while the weight of "complexity of the current profile cross-sectional geometry" is appropriately increased. The weight of "fluctuation information of extrusion speed under stable operating conditions" is adjusted according to its correlation with the pattern. Through this dynamic weight adjustment, the resulting risk metric more accurately reflects the risk of this specific deep stress accumulation pattern under current production conditions, thereby guiding subsequent process parameter adjustments. These adjustments could include increasing local cooling intensity at the flange, optimizing die design to reduce wall thickness variations, or adjusting the stability of the extrusion speed to effectively suppress the generation of deep stress.

[0077] In the aforementioned intelligent control method for aluminum profile production and processing, production process parameters are adjusted based on comprehensive evaluation results, including: S301. Obtain the alloy grade, cross-sectional geometry, target product mechanical performance requirements, and expected production cycle of the profile.

[0078] Specifically, this information forms the basis for establishing and applying a set of nonlinear relational functions, which define the boundary conditions and performance targets of the production process. For example, different alloy grades have different sensitivities to cooling rates, complex cross-sectional geometries can lead to uneven cooling, and the target mechanical property requirements and expected production cycle time directly affect the range of cooling intensity and extrusion speed settings.

[0079] S302. Based on the alloy grade, cross-sectional geometry, mechanical performance requirements of the target product, and expected production cycle, extract a set of nonlinear relationship functions describing the effects of cooling intensity and extrusion speed on the internal stress accumulation of the profile and the energy consumption of the production process.

[0080] This set of functions can be understood as a set of mathematical models or empirical formulas used to quantify how the two key process parameters, cooling intensity and extrusion speed, nonlinearly affect the risk measure of internal stress accumulation in the profile and the energy consumption of the production process. These nonlinear relationship functions can be obtained through historical production data analysis, finite element simulation, machine learning model training, etc., with the aim of accurately capturing the complex relationship between parameter changes and result responses.

[0081] S303. A transmission protocol based on time synchronization and data packet sequence number is used to transmit redundant sampled data.

[0082] The time synchronization mechanism ensures the consistency of data collected by different sensors along the timeline, which is crucial for analyzing the dynamic behavior of battery cells. The data packet sequence number is used to detect data packet loss or out-of-order delivery, enabling accurate reconstruction of the original data stream at the receiving end.

[0083] S304. Substitute the risk measurement and energy consumption information into a set of nonlinear relational functions to evaluate the comprehensive impact of the current combination of production process parameters on the internal stress accumulation of the profile and the energy consumption of the production process, and obtain the evaluation results.

[0084] This assessment reflects the risk level and energy consumption level of internal stress accumulation in the profile under the current process parameters.

[0085] S305. Based on the evaluation results, adjust the cooling intensity and extrusion speed until a combination of production process parameters is found that achieves a global balance between the risk measure of internal stress accumulation in the profile and the energy consumption of the production process.

[0086] The adjustment process can employ optimization algorithms (such as gradient descent, genetic algorithms, or particle swarm optimization) to gradually approach the optimal solution by continuously exploring and evaluating different parameter combinations. The goal of the adjustment is to find a combination of production process parameters that achieves a global balance between the risk measure of internal stress accumulation in the profile and the energy consumption of the production process. Global balance means that, while meeting the requirements for profile surface quality and immediate performance, both risk measure and energy consumption are at an acceptable minimum level, and production efficiency is optimal.

[0087] This application's solution, by introducing a set of nonlinear relational functions, can more accurately capture the complex impact of key process parameters such as cooling intensity and extrusion speed on the internal stress accumulation and production energy consumption of profiles. Traditional methods may rely solely on experience or simple linear models for parameter adjustment, which is insufficient to handle the multi-factor, nonlinear coupling effects in actual production, resulting in low adjustment efficiency or failure to achieve a true optimal balance. By obtaining detailed information such as the profile's alloy grade, cross-sectional geometry, target product mechanical performance requirements, and expected production cycle time, precise boundary conditions and target constraints can be provided for the construction and application of the nonlinear relational function set, making the model more targeted and accurate. Substituting real-time risk metrics and energy consumption information into these nonlinear functions for evaluation can quantify the comprehensive impact of the current combination of process parameters. Based on this, by adjusting the cooling intensity and extrusion speed through iterative optimization algorithms, the system can intelligently explore the parameter space until it finds a combination of production process parameters that achieves a global balance between the risk metrics of internal stress accumulation and the energy consumption of the production process while ensuring product quality and performance. This model-based optimization method significantly improves the efficiency and accuracy of parameter adjustment, avoids blind trial and error, and effectively solves the problems of inaccurate adjustment and poor optimization effect that may exist in the basic scheme.

[0088] In some preferred embodiments, a specific example is given below. Assume the production of an aluminum profile with a specific alloy grade of 6063, featuring a complex hollow, irregular cross-sectional geometry. The target product requires high tensile and yield strength, and the expected production cycle is 10 meters per minute. First, the system acquires this detailed production information. Then, based on historical production data and finite element simulation results, the system extracts a set of nonlinear relational functions. This set describes the variation of risk measures (e.g., residual stress level) of internal stress accumulation and energy consumption (e.g., electrical energy consumption per unit length of profile) in the production process under different cooling intensities (e.g., cooling fan speed, cooling water flow rate) and extrusion speeds. For example, one function might represent the nonlinear relationship between cooling intensity and surface temperature gradient, while another might represent the nonlinear relationship between extrusion speed and internal heat accumulation. During production, once the risk measures and energy consumption information are acquired in real time, they are substituted into this set of nonlinear relational functions to assess the combined impact of the current combination of cooling intensity and extrusion speed. For example, if the assessment results show that the current risk metric is high while energy consumption is low, the system will predict using a nonlinear function and then fine-tune the cooling intensity and extrusion speed. For instance, it might appropriately reduce the extrusion speed and increase the cooling intensity to reduce the risk without significantly increasing energy consumption. This adjustment process is iterative; the system will continuously evaluate and adjust the parameters until it finds an optimal combination of cooling intensity and extrusion speed. This results in a state where, while meeting the product's mechanical performance requirements and production cycle time, the risk metric of internal stress accumulation in the profile and production energy consumption are at their lowest globally and in a balanced state.

[0089] In some embodiments described above in this application, by adjusting the cooling intensity and extrusion speed, the aim is to find a combination of production process parameters that achieves a global balance between the risk measure of internal stress accumulation in the profile and the energy consumption of the production process. However, in actual optimization processes, especially when facing complex nonlinear relationships, traditional iterative adjustment methods may risk getting trapped in local optima, resulting in the inability to find the true global optimum, thereby affecting the final balance effect and production efficiency.

[0090] In this regard, this application further proposes that the above-mentioned method for adjusting production process parameters also includes: S401 Record the current iteration's combination of process parameters, risk measurement of internal stress accumulation in the profile, and energy consumption information of the production process.

[0091] Specifically, after each iteration and adjustment, the system stores the current combination of process parameters such as cooling intensity and extrusion speed, as well as the risk measurement and production energy consumption data of the resulting internal stress accumulation of the profile.

[0092] S402. Compare the current combination of process parameters with the local optimal solutions on the historical optimization path to determine whether the optimization process has fallen into a local state.

[0093] Specifically, this can be determined by analyzing the changing trends of risk metrics and energy consumption information in historical records. For example, if the changes in risk metrics and energy consumption information are all less than a preset threshold in multiple consecutive iterations, or if similar parameter combinations and evaluation results repeatedly appear in a certain region, then the optimization process may have fallen into a local optimum.

[0094] S403. When the optimization process gets stuck in a local state, adjust the set of nonlinear relational functions.

[0095] When the system identifies itself trapped in a local optimum, it no longer merely fine-tunes process parameters, but modifies the set of nonlinear relational functions used to describe the impact of cooling intensity and extrusion speed on the accumulation of internal stress in the profile and the energy consumption of the production process. This adjustment can include changing the parameters of the function model, introducing new variables, or switching to a different function model to alter the characteristics of the optimization space, thereby helping to escape local optima.

[0096] Specifically, when the optimization process gets stuck in a local state, the fine-tuning step size can be increased; based on the adjusted fine-tuning step size, the key parameters in the set of nonlinear relational functions can be fine-tuned.

[0097] Specifically, "increasing the fine-tuning step size" refers to intentionally increasing the step size used to adjust the parameters in the set of nonlinear relational functions during the optimization algorithm. This fine-tuning step size can be understood as the magnitude of parameter updates, aiming to help the algorithm escape the current local optimum by making larger parameter changes when the optimization process gets stuck in a local state. For example, the original fine-tuning step size can be multiplied by a coefficient greater than 1, or a preset large step size value can be set directly. "Key parameters in the set of nonlinear relational functions" refer to the parameters that have the most significant impact on the risk measurement of internal stress accumulation in the profile and the energy consumption of the production process. These key parameters typically include, but are not limited to, the cooling coefficient in the cooling intensity function and the extrusion modulus in the extrusion speed function. Fine-tuning these key parameters means concentrating resources on adjusting the parameters with the greatest impact while keeping other parameters relatively stable, in order to achieve a more efficient optimization effect.

[0098] S404. Based on the adjusted set of nonlinear relationship functions, re-evaluate and adjust until a combination of production process parameters is found that can achieve a global balance between the risk measure of internal stress accumulation in the profile and the energy consumption of the production process.

[0099] Specifically, after the set of nonlinear relational functions is adjusted, the system will re-evaluate the combined impact of the current combination of process parameters on risk measurement and energy consumption based on the new set of functions, and then iteratively adjust again. This process will continue until a combination of production process parameters that achieves a global balance between risk measurement and energy consumption is found.

[0100] This application's solution effectively addresses the problem of traditional optimization methods potentially getting trapped in local optima by introducing a mechanism for judging the local state of the optimization process and dynamically adjusting the set of nonlinear relational functions based on this mechanism. Specifically, by recording and comparing historical optimization data, the system can identify whether the optimization trajectory has stagnated or is looping within a local region, thus determining whether it has fallen into a local state. Once a local state is identified, the system is no longer limited to fine-tuning process parameters but fundamentally changes the behavior of the optimization model and the optimization space by adjusting the set of nonlinear relational functions. This allows the optimization algorithm to escape the current local optimum trap, explore a broader parameter space, and thus is more likely to find the true global optimum. This strategic adjustment ensures a global balance between production efficiency and the risk measurement of internal stress accumulation in profiles, even in complex and ever-changing environments.

[0101] Through the above technical solution, this application effectively avoids the optimization process from getting trapped in local optima, significantly improving the global convergence capability of the optimization algorithm. Compared with traditional methods that rely solely on parameter fine-tuning, this application dynamically adjusts the set of nonlinear relational functions, making the optimization process more adaptable and robust. It can more accurately and efficiently find the global balance point between production efficiency and the risk measure of internal stress accumulation in the profile. Therefore, it not only further improves the production quality and stability of aluminum profiles but also maximizes production efficiency and reduces energy consumption, providing a more reliable and advanced solution for the intelligent control of aluminum profile production and processing.

[0102] In some preferred embodiments, it is assumed that during the aluminum profile production process, the system optimizes the cooling intensity and extrusion speed using the aforementioned method of adjusting production process parameters. After hundreds of iterations, the system finds that the risk measurement of internal stress accumulation and the assessment results of production energy consumption fluctuate within a narrow range, and the combination of process parameters tends to stabilize, but has not reached the expected optimal balance point. At this point, the system determines that the optimization process has fallen into a local state. To escape this local state, the system adjusts the set of nonlinear relational functions describing the impact of cooling intensity and extrusion speed on risk measurement and energy consumption. For example, a random perturbation term can be introduced into the function model, or the weight coefficients in the function can be adjusted, or even a more exploratory function form can be switched. After adjustment, the system restarts the evaluation and adjustment process using the new set of nonlinear relational functions. In this way, the optimization algorithm can explore parameter regions that were not previously fully considered, and ultimately successfully find a better combination of process parameters, so that the risk measurement of internal stress accumulation and production energy consumption of the profile reach a global balance, thereby achieving higher production efficiency and more stable product quality.

[0103] In some embodiments described above, this application proposes a scheme to adjust the set of nonlinear relation functions to escape local optima when the optimization process gets stuck in a local state. However, in actual implementation, local states may take various forms, such as saddle point regions or flat regions. If these different types of local states are not distinguished and targeted adjustment strategies are not adopted, the adjustment efficiency may be low, or even unable to effectively escape local optima, thus affecting the efficiency and accuracy of finding the global optimum. Therefore, this application further proposes a more refined method for adjusting the set of nonlinear relation functions, which identifies the specific type of local state and adopts differentiated strategies to optimize the optimization process.

[0104] When the optimization process gets stuck in a local state, the set of nonlinear relational functions is adjusted, specifically including: S501. Obtain the combination of process parameters for the current optimization iteration.

[0105] For example, it can be a set of parameters such as cooling intensity and extrusion speed determined by the current optimization algorithm.

[0106] S502. Based on the current combination of process parameters, risk measurement, and energy consumption information in the optimization iteration, calculate the curvature change and gradient information of the optimization trajectory.

[0107] Specifically, curvature changes reflect the degree of bending of the optimization path, while gradient information indicates the direction and rate of change of the objective function value in the parameter space. This information is used to quantify the local terrain features around the current optimization point.

[0108] S503. Based on the curvature change and gradient information of the optimization trajectory, determine whether the optimization process has entered the saddle point region or the flat region.

[0109] A saddle point is a point where the objective function value increases in some directions and decreases in others. Its characteristic is that the gradient is close to zero, but the Hessian matrix has positive and negative eigenvalues. A flat region, on the other hand, refers to a region where the objective function value changes very little over a large range, and the gradient is also close to zero. By analyzing curvature and gradient, these two local states can be accurately distinguished.

[0110] S504. When the optimization process enters the saddle point region, introduce random disturbances to change the current combination of process parameters and escape the saddle point region.

[0111] Random perturbation refers to adding a small random quantity to the current combination of process parameters. Its purpose is to break the equilibrium state of the saddle point and enable the optimization process to continue along the direction of the non-zero gradient.

[0112] S505. When the optimization process is determined to have entered a flat region, increase the exploratory step size to expand the parameter search range.

[0113] Exploratory step size refers to using a larger step size when updating parameters. Its purpose is to quickly traverse flat regions and avoid consuming too many computing resources in locally flat regions, thereby accelerating the optimization process.

[0114] S506. Adjust the set of nonlinear relational functions based on the changed combination of current process parameters or the combination of parameters after expanding the parameter search range.

[0115] This application's solution, by introducing the calculation of curvature changes and gradient information of the optimization trajectory, can accurately identify whether the local state trapped in the optimization process is a saddle point region or a flat region. It is precisely this refined identification of local states that allows this application to adopt different strategies to adjust the set of nonlinear relational functions. Specifically, when the optimization process is identified as trapped in a saddle point region, introducing random perturbations can effectively break the equilibrium at the saddle point, allowing the optimization process to continue exploring in a new direction, thus avoiding prolonged stagnation at the saddle point. Conversely, when the optimization process is identified as trapped in a flat region, increasing the exploratory step size allows the optimization algorithm to quickly traverse these regions with insignificant changes, avoiding getting trapped in local optima or having excessively slow convergence. This differentiated approach enables the optimization algorithm to adopt the most effective escape strategy when facing different types of local traps, thereby significantly improving the optimization efficiency and the ability to discover the global optimum.

[0116] In some preferred embodiments, it is assumed that during the aluminum profile manufacturing process, the intelligent control system is adjusting the cooling intensity and extrusion speed to balance the risk metric of internal stress accumulation and the energy consumption of the production process. In a certain iteration, the system finds that the change in the objective function (comprehensive evaluation result) is very small, with the gradient close to zero. At this point, the system calculates the curvature change and gradient information of the optimization trajectory. If the analysis results show that, in the parameter space of cooling intensity and extrusion speed, the objective function value at the current point slightly increases in one direction and slightly decreases in another, and the eigenvalues ​​of the Hessian matrix are both positive and negative, then it is determined that the system is currently in a saddle point region. To escape this saddle point, the system adds a random perturbation within a preset range to the current cooling intensity and extrusion speed parameters; for example, fine-tuning the cooling intensity by +0.5% and the extrusion speed by -0.2%. Through this random perturbation, the new parameter combination will allow the system to escape the saddle point and continue exploring a better solution.

[0117] On the other hand, if the analysis results show that the objective function values ​​around the current point hardly change over a large range in the parameter space of cooling intensity and extrusion speed, the gradient is close to zero, and the eigenvalues ​​of the Hessian matrix are all close to zero, then it is determined that the current point is in a flat region. To quickly traverse this flat region, the system increases the exploratory step size, for example, doubling the adjustment step size for cooling intensity and extrusion speed, thereby conducting a bolder exploration in the parameter space to quickly find the region where the objective function values ​​begin to change significantly, thus accelerating the optimization process and ultimately finding a combination of production process parameters that achieves a global balance between the risk measure of internal stress accumulation in the profile and the energy consumption of the production process.

[0118] In some embodiments described above in this application, when the optimization process gets stuck in a local state, the local optimum can be escaped by adjusting the set of nonlinear relational functions. One way to do this is to introduce random perturbations to change the current combination of process parameters when the optimization process enters a saddle point region. However, if the introduction of random perturbations lacks necessary constraints and prediction mechanisms, it may cause the process parameters to deviate from the safe production range, thereby affecting the surface quality and immediate performance of the profile, and even causing production instability. If the above problems are not addressed, product quality may be sacrificed or production risks may increase in the pursuit of the global optimum.

[0119] In response, this application further proposes a step to introduce random perturbations when the optimization process enters the saddle point region, thereby changing the current combination of process parameters and escaping the saddle point region. This step includes: S601. Obtain the alloy grade, cross-sectional geometry, and mechanical performance requirements of the target product for the profile.

[0120] Among these factors, the alloy grade determines the basic physical and chemical properties of the material, the cross-sectional geometry affects the uniformity of heat exchange, and the mechanical performance requirements of the target product set the quality standards for the final product. This information forms the basis for subsequently determining the safety fluctuation boundary.

[0121] S602. Determine the instantaneous safety fluctuation boundaries of cooling intensity and extrusion speed based on the alloy grade, cross-sectional geometry, and mechanical property requirements of the target product.

[0122] The instantaneous safety fluctuation boundary can be understood as the maximum allowable instantaneous variation range of cooling intensity and extrusion speed parameters without compromising the surface quality and immediate performance of the profile. This boundary can be determined based on a pre-established material database, finite element simulation models, or historical production data analysis. For example, for a specific alloy and section, excessively high cooling intensity may lead to surface cracks, while excessively low extrusion speed may affect production efficiency; therefore, a dynamic safety range needs to be set.

[0123] S603. Within the instantaneous safety fluctuation boundary, generate a random disturbance.

[0124] This random perturbation is a value randomly generated within a safety boundary, used to fine-tune the current cooling intensity and extrusion speed parameters. It can be generated using a uniform distribution, Gaussian distribution, or other random number generation algorithms that conform to a specific strategy, aiming to provide diverse exploration directions for the optimization process while ensuring safety.

[0125] S604. The random disturbance is superimposed on the current cooling intensity and extrusion speed parameters to form an exploratory parameter combination.

[0126] This means adding or subtracting the currently used cooling intensity and extrusion speed parameters from the generated random disturbance to obtain a new set of process parameters to be verified.

[0127] S605, the influence of the combination of preliminary experimental parameters on the surface temperature distribution at the outlet of the profile cooling zone, the instantaneous tensile strength and yield strength of the profile.

[0128] Specifically, the prediction process can utilize pre-trained machine learning models, physical simulation models, or regression models based on historical data. Its purpose is to assess whether new parameter combinations will lead to abnormal surface temperature distribution of the profile (such as localized overcooling or overheating) and whether they will cause the profile's instantaneous tensile strength and yield strength to fall below target requirements before actual application on the production line.

[0129] S606. When the prediction results show that the cooling intensity and extrusion speed are within the instantaneous safety fluctuation boundary, apply the trial parameter combination to the production line to change the current process parameter combination and get out of the saddle point region.

[0130] When the forecast results show that the cooling intensity and extrusion speed are within the instantaneous safety fluctuation boundaries, a trial parameter combination is applied to the production line to change the current process parameter combination and escape the saddle point region. This means that the new parameter combination will only be implemented in actual production if the forecast results confirm that it will not negatively impact product quality and production stability. If the forecast results do not meet the conditions, the disturbance will be regenerated or other strategies will be adopted.

[0131] The proposed solution acquires key production information of the profile before introducing random disturbances when the optimization process enters the saddle point region. Based on this information, it dynamically determines the instantaneous safety fluctuation boundaries of cooling intensity and extrusion speed. By generating random disturbances within these safety boundaries and predictively evaluating the resulting tentative parameter combinations, the introduced disturbances can effectively alter the current process parameter combinations without compromising the profile's surface quality and immediate performance. Thus, the optimization process can successfully escape the locally optimal saddle point region and continue converging towards the globally optimal solution while ensuring production safety and product quality.

[0132] By employing the aforementioned technical solution, when the optimization process becomes trapped in a saddle point region, random disturbances can be introduced in a controlled and safe manner, avoiding production risks and quality problems that may arise from blind disturbances. This solution ensures that while exploring better combinations of process parameters, the surface quality and real-time performance of the profiles always meet requirements, thereby improving the robustness and reliability of the intelligent control method, accelerating the discovery of the global optimum, and effectively reducing potential risks in the production process.

[0133] In some preferred embodiments, it is assumed that the aluminum profile currently being produced is made of 6063 alloy with a complex irregular cross-sectional geometry, and the target product requires a tensile strength of not less than 200 MPa and a yield strength of not less than 160 MPa. The system first calculates the instantaneous safe fluctuation boundaries of the current cooling intensity and extrusion speed based on this information, combined with historical data and a material property model. For example, the cooling intensity is allowed to fluctuate within ±5% of the current value, and the extrusion speed is allowed to fluctuate within ±3% of the current value. Subsequently, the system generates a random perturbation within these boundaries, such as increasing the cooling intensity by 2% and decreasing the extrusion speed by 1%. This perturbation is superimposed on the current process parameters to form an exploratory parameter combination. Next, a pre-trained neural network model is used to predict the impact of this exploratory parameter combination on the surface temperature distribution at the outlet of the profile cooling zone (e.g., predicting a maximum temperature difference not exceeding 5°C) and the instantaneous tensile strength and yield strength. If the prediction results show that all indicators are within the safe range, and the instantaneous tensile strength and yield strength are still higher than the target requirements, then the trial parameter combination is applied to the production line, so that the optimization process can break out of the current saddle point area and continue to find a better balance point.

[0134] like Figure 3 As shown in the figure, this invention also provides an intelligent control system for the aluminum profile manufacturing process. The system includes: The parameter acquisition module is used to acquire various production process parameters related to the internal stress accumulation of the profile. These parameters include the surface temperature distribution of the profile at the cooling zone outlet, the difference between the actual operating power and the theoretical required power of the cooling fan, the complexity of the cross-sectional geometry of the current profile, and the fluctuation of the extrusion speed under stable operating conditions. The risk measurement module is used to measure the risk of internal stress accumulation in profiles based on various production process parameters. The comprehensive assessment module is used to generate comprehensive assessment results based on risk measurement and energy consumption information of the production process. The comprehensive assessment results are used to evaluate the relationship between quantitative production efficiency and risk measurement. The parameter adjustment module is used to adjust the production process parameters based on the comprehensive evaluation results, so as to achieve a balance between production efficiency and risk measurement while ensuring that the surface quality and immediate performance of the profiles meet the requirements.

[0135] This application also provides a computer-readable storage medium. All or part of the processes in the above method embodiments can be executed by a computer program instructing related hardware. This program can be stored in the computer-readable storage medium, and when executed, it can include the processes of the above method embodiments. The computer-readable storage medium can be an internal storage unit of the task execution device (including a data sending end and / or a data receiving end) of any of the foregoing embodiments, such as the hard disk or memory of the task execution device. The computer-readable storage medium can also be an external storage device of the terminal device, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the terminal device. Further, the computer-readable storage medium can include both the internal storage unit of the task execution device and an external storage device. The computer-readable storage medium is used to store the computer program and other programs and data required by the task execution device. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0136] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0137] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, essentially, or the parts that contribute to the prior art, or all or part of the technical solutions, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0138] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be covered within the scope of protection of this application.

Claims

1. A method for intelligent control of aluminum profile manufacturing process, characterized in that, include: Acquire various production process parameters related to the accumulation of internal stress in the profile; these parameters include the surface temperature distribution of the profile at the outlet of the cooling zone, the difference between the actual operating power and the theoretical required power of the cooling fan, the complexity of the cross-sectional geometry of the currently produced profile, and the fluctuation of the extrusion speed under stable operating conditions. Based on the aforementioned various production process parameters, a risk measure for internal stress accumulation in the profile is obtained; Based on the risk measurement and energy consumption information of the production process, a comprehensive evaluation result is generated. The comprehensive evaluation result is used to evaluate the relationship between quantitative production efficiency and the risk measurement. Based on the comprehensive evaluation results, the production process parameters are adjusted to achieve a balance between production efficiency and the risk metric while ensuring that the surface quality and immediate performance of the profiles meet the requirements.

2. The intelligent control method for aluminum profile production and processing according to claim 1, characterized in that, The risk metric for internal stress accumulation in the profile, obtained based on the various production process parameters, includes: Based on the various production process parameters and their corresponding mapping relationships, determine the risk value corresponding to each of the production process parameters; Based on the weight corresponding to each of the production process parameters, the risk values ​​corresponding to each of the production process parameters are weighted and summed to obtain a risk measure of internal stress accumulation in the profile.

3. The intelligent control method for aluminum profile production and processing according to claim 2, characterized in that, The step of weighted summation of risk values ​​corresponding to each of the production process parameters based on their respective weights to obtain a risk measure for internal stress accumulation in the profile includes: Based on the weight corresponding to each of the production process parameters, the risk values ​​corresponding to each of the production process parameters are weighted and summed to obtain the initial risk measure of internal stress accumulation in the profile. The cross-section of the profile is divided into multiple micro-regions, and the geometric thermal sensitivity index of each micro-region is calculated; the geometric thermal sensitivity index is used to reflect the tendency of the micro-region to induce deep stress when the cooling is uneven. For any of the aforementioned micro-regions, when the deviation ratio between the actual temperature decrease rate or temperature fluctuation frequency of the micro-region and the expected dynamic cooling trajectory is greater than a deviation threshold, and the deviation exhibits non-periodic characteristics, a transient heat exchange anomaly is identified in the micro-region; and The initial risk metric is corrected based on the geometric thermal sensitivity index of regions with transient heat exchange anomalies to obtain the risk metric.

4. The intelligent control method for aluminum profile production and processing according to claim 2, characterized in that, The method further includes: The surface temperature distribution information is analyzed for local temperature gradient and high-frequency fluctuation characteristics are extracted. The difference between the actual operating power and the theoretical required power of the cooling fan is analyzed for trend and the duration of abnormality is assessed. The complexity of the cross-sectional geometry of the current production profile is quantified by wall thickness difference and irregularity index. The fluctuation information of the extrusion speed under stable operating conditions is analyzed by spectrum analysis to obtain the parameter characteristics of the production process parameters. The irregularity index is the area of ​​the cross-section and the minimum circumscribed rectangle or circle. Multi-dimensional correlation analysis is performed on the parameter characteristics to identify deep stress accumulation modes caused by the synergistic effect of multiple parameters; Based on the deep stress accumulation mode, the initial weights corresponding to each of the production process parameters are adjusted to obtain the weights corresponding to each of the production process parameters.

5. The intelligent control method for aluminum profile production and processing according to claim 1, characterized in that, The step of adjusting production process parameters based on the comprehensive evaluation results includes: Obtain the alloy grade, cross-sectional geometry, mechanical performance requirements of the target product, and expected production cycle of the profile; Based on the alloy grade of the profile, the cross-sectional geometry, the mechanical performance requirements of the target product, and the expected production cycle, a set of nonlinear relational functions describing the effects of cooling intensity and extrusion speed on the internal stress accumulation of the profile and the energy consumption of the production process is extracted. Substituting the risk metric and energy consumption information into the set of nonlinear relational functions, the combined impact of the current combination of production process parameters on the internal stress accumulation of the profile and the energy consumption of the production process is evaluated to obtain the evaluation results. Based on the evaluation results, the cooling intensity and extrusion speed are adjusted until a combination of production process parameters is found that achieves a global balance between the risk measure of internal stress accumulation in the profile and the energy consumption of the production process.

6. The intelligent control method for aluminum profile production and processing according to claim 5, characterized in that, The method further includes: Record the current combination of process parameters, risk measurement of internal stress accumulation in the profile, and energy consumption information of the production process; Compare the current combination of process parameters with the local optimal solutions on the historical optimization path to determine whether the optimization process has fallen into a local state; When the optimization process gets stuck in a local state, the set of nonlinear relational functions is adjusted; Based on the adjusted set of nonlinear relational functions, the evaluation and adjustment are carried out again until a combination of production process parameters that can achieve a global balance between the risk measure of internal stress accumulation in the profile and the energy consumption of the production process is found.

7. The intelligent control method for aluminum profile production and processing according to claim 6, characterized in that, When the optimization process gets stuck in a local state, adjusting the set of nonlinear relation functions includes: When the optimization process gets stuck in a local state, increase the fine-tuning step size; Based on the adjusted fine-tuning step size, the key parameters in the set of nonlinear relational functions are fine-tuned.

8. The intelligent control method for aluminum profile production and processing according to claim 6, characterized in that, When the optimization process gets stuck in a local state, adjusting the set of nonlinear relation functions includes: Obtain the current combination of process parameters for the optimization iteration; Based on the current combination of process parameters in the optimization iteration, the risk metric, and the energy consumption information, calculate the curvature change and gradient information of the optimization trajectory; Based on the curvature change and gradient information of the optimization trajectory, determine whether the optimization process has entered a saddle point region or a flat region. When the optimization process enters the saddle point region, a random disturbance is introduced to change the current combination of process parameters and escape the saddle point region. When the optimization process is determined to have entered the flat region, the exploratory step size is increased to expand the parameter search range; The set of nonlinear relational functions is adjusted based on the changed combination of current process parameters or the combination of parameters after expanding the parameter search range.

9. The intelligent control method for aluminum profile production and processing according to claim 8, characterized in that, When the optimization process enters the saddle point region, a random disturbance is introduced to change the current combination of process parameters and escape the saddle point region, including: Obtain the alloy grade, cross-sectional geometry, and mechanical property requirements of the target product for the profile; Based on the alloy grade, the cross-sectional geometry, and the mechanical property requirements of the target product, determine the instantaneous safety fluctuation boundaries of cooling intensity and extrusion speed; Within the instantaneous safety fluctuation boundary, a random disturbance is generated; The random disturbance is superimposed on the current cooling intensity and extrusion speed parameters to form an exploratory parameter combination; Predict the effects of the proposed combination of parameters on the surface temperature distribution at the outlet of the profile cooling zone, the instantaneous tensile strength and yield strength of the profile; When the prediction results show that the cooling intensity and extrusion speed are within the instantaneous safety fluctuation boundary, the tentative parameter combination is applied to the production line to change the current process parameter combination and break out of the saddle point region.

10. An intelligent control system for aluminum profile manufacturing and processing, characterized in that, The system includes: The parameter acquisition module is used to acquire various production process parameters related to the internal stress accumulation of the profile. These various production process parameters include the surface temperature distribution information of the profile at the outlet of the cooling zone, the difference between the actual operating power and the theoretical required power of the cooling fan, the complexity of the cross-sectional geometry of the current profile, and the fluctuation information of the extrusion speed under stable operating conditions. The risk measurement module is used to obtain a risk measurement of the internal stress accumulation of the profile based on the various production process parameters. The comprehensive evaluation module is used to generate a comprehensive evaluation result based on the risk measurement and energy consumption information of the production process. The comprehensive evaluation result is used to evaluate the relationship between quantitative production efficiency and the risk measurement. The parameter adjustment module is used to adjust the production process parameters based on the comprehensive evaluation results, so as to achieve a balance between production efficiency and the risk measurement while ensuring that the surface quality and immediate performance of the profile meet the requirements.

Images