Extrusion cooling online detection and feedback correction system

Abstract

The invention relates to the field of aluminum alloy profile extrusion cooling, and discloses an extrusion cooling online detection and feedback correction system which is characterized by comprising a parameter acquisition module, a cooling data acquisition module, a cooling device calibration module, a data processing and control module and a cooling device controller. By organically combining multi-parameter sensing, device precise calibration, multi-dimensional state evaluation, intelligent function matching, closed-loop feedback adjustment and machine learning self-optimization, a highly-intelligent, self-adaptive and high-precision cooling control system is constructed, the industrial pain point in the aluminum alloy extrusion cooling process is fundamentally solved, and the production efficiency is improved. And the method has great value for improving the product quality and the production efficiency of the high-end aluminum alloy profile.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of aluminum alloy profile extrusion cooling, and more particularly to an extrusion cooling online detection and feedback correction system. BACKGROUND

[0002] Aluminum alloy extrusion cooling refers to a process step of forcibly cooling high-temperature profiles such as rods, tubes, and sections that have just been extruded from a die during the aluminum alloy extrusion forming process. This is a crucial link in the aluminum alloy extrusion production line, directly affecting the microstructure, mechanical properties, dimensional accuracy, surface quality, and production efficiency of the final product. With the widespread application of aluminum alloy profiles in the industrial field, the importance of the cooling link in the production process to the quality and structural stability of the finished product has become increasingly prominent. Traditional cooling control systems mainly rely on experience parameter setting and manual adjustment, usually only providing fixed air cooling or water cooling schemes, and the lack of adaptability problem gradually emerges. This way may cause overcooling, undercooling, or uneven cooling due to the mismatch between cooling intensity and profile thermal state, thereby affecting the hardness, dimensional accuracy, and surface quality of the profile.

[0003] Some improved schemes introduce cooling mode selection and sensor collection functions, but they are still at the stage of quantitatively supplying wind / water according to regions, and have not yet realized dynamic feedback closed-loop adjustment of the cooling process. In addition, the prediction and control ability of the cooling response lag phenomenon still needs to be improved, which makes the adjustment lag and overshoot phenomenon common. SUMMARY

[0004] In view of the deficiencies of the prior art, the purpose of the present application is to provide an extrusion cooling online detection and feedback correction system to overcome the above-mentioned defects in the prior art, and to solve the problems of uneven cooling, insufficient hardness, dimensional accuracy deviation, and surface quality caused by the lack of self-adaptive control ability in the cooling link of the aluminum alloy profile extrusion production process in the prior art.

[0005] To achieve the above-mentioned purpose, the present application provides the following technical solutions: An extrusion cooling online detection and feedback correction system comprises a parameter acquisition module for acquiring physical parameters of the metal profile to be cooled, the physical parameters including material thermal conductivity, specific heat capacity, cross-sectional structure, initial temperature, target cooling temperature, and cooling time requirement, the cooling time requirement including total time of the cooling process and time distribution requirement of the cooling rate; a cooling data acquisition module for acquiring cooling data in the cooling process, the cooling data including temperature fluctuation data, thermal distribution data, and heat conduction data; The cooling device calibration module is configured to test output characteristics of the liquid cooling device, the fog cooling device, and the air cooling device, and to establish a correspondence model between the control instructions and actual output parameters of the cooling device. The data processing and control module is configured to calculate cooling stability, thermal distribution uniformity, and thermal conductivity based on cooling data, to construct a multi-dimensional cooling state vector, to match the real-time updated multi-dimensional cooling state vector with an execution function in a preset cooling control function library, and to generate a control instruction. The cooling device controller is configured to receive the control instruction to adjust operating parameters of the cooling device.

[0006] Further, the cooling device calibration module comprises: The liquid cooling calibration unit is configured to establish a correction relationship model between liquid cooling device control instructions and cooling liquid flow and temperature output by least square fitting; The air cooling calibration unit is configured to establish a mapping relationship model between air pressure set values of the fog cooling device and air flow speed and coverage range by regression analysis; The air cooling calibration unit is configured to establish a mapping relationship model between air pressure set values of the fog cooling device and air flow speed and coverage range by regression analysis; The calibration database is configured to store the correction relationship model, the mapping relationship model, and the correlation model.

[0007] Further, the data processing and control module comprises: The data preprocessing unit is configured to filter and remove outliers from the cooling data; The vector calculation unit is configured to calculate index values of each dimension of the vector based on a preset quantization formula; wherein, The cooling stability is quantified by a temperature fluctuation coefficient, and the calculation formula is the ratio of the temperature standard deviation to the average temperature; The thermal distribution uniformity is quantified by a temperature distribution uniformity, and the calculation formula is 1 minus the ratio of the temperature sample variance to the overall variance. Further, the data processing and control module further comprises a function matching unit, which is configured to: Read the latest multi-dimensional cooling state vector and calculate its Euclidean distance with the previous period vector; If the Euclidean distance exceeds a preset threshold, or the latest vector exceeds the vector interval corresponding to the current execution function, the matching process is started; A matching strategy based on a preset vector interval is used to screen candidate execution functions from the cooling control function library, and the cosine similarity between the multi-dimensional cooling state vector and the preset vector of each candidate function is calculated; Select the cosine similarity of the highest execution function as the target function, and combine the model in the cooling device calibration module to convert the output parameters of the target function into control instructions. Further, it also includes a feedback and optimization module, which is configured to: Continuously collect cooling data and calculate a multi-dimensional cooling state vector, and compare it with an ideal cooling vector, which is a target state vector calculated according to cooling target parameters; Calculate the Euclidean distance between the two as a cooling effect difference index; If the difference index is within the preset allowable deviation threshold range, fine-tune and optimize the parameters of the current execution function; If the difference index exceeds the preset allowable deviation threshold range, trigger the data processing and control module to perform function matching again; If it is monitored that the cooling device operating parameters continuously deviate from the standard range, an alarm signal is generated and a preset emergency cooling function is called. Further, the feedback and optimization module also includes a fault diagnosis unit, which is configured to: Monitor the cooling device operating parameters and compare them with the preset standard range; If the parameters deviate, the corresponding model in the calibration database is called to calculate compensation parameters, which are used to correct the operating parameters; If the compensated operating parameters still exceed the preset allowable deviation threshold range, it is confirmed as a fault state, a fault alarm signal is generated and recorded to the log database. Further, it also includes a function library optimization module, which is configured to: Analyze the profile parameters, multi-dimensional cooling state vectors, execution function selection records and cooling effect evaluation results in the historical cooling process; Statistical each cooling function in different vector intervals matching success rate, the matching success rate is defined as the proportion of function output meeting the target state vector requirements; Identify the vector interval with a matching success rate below the preset threshold, and adjust the parameters of the cooling function or add new adaptive functions in this interval. Further, a profile deformation amount dimension is added to the multi-dimensional cooling state vector, and its quantitative formula is the ratio of the actual deformation amount of the profile to the target deformation amount, which is obtained by measuring the deformation monitoring device; Based on experimental test data, verify the contribution of the profile deformation amount dimension, and adjust the weight distribution of the multi-dimensional cooling state vector; Update the vector database and vector-function mapping model to adapt to the expanded multi-dimensional cooling state vector space.

[0008] Further, the function library optimization module adopts a machine learning model to optimize the vector-function mapping relationship, and the machine learning model is a support vector machine or a decision tree model. The machine learning model takes a historical multi-dimensional cooling state vector as input and executes function selection records as output for training. The model performs performance evaluation through cross-validation and is deployed in the system after the accuracy reaches a preset threshold.

[0009] Further, a cooling dynamic optimization module is further included, and the cooling dynamic optimization module is configured to: The cooling data is continuously collected, and cooling stability, thermal distribution uniformity, thermal conductivity coefficient and profile deformation amount dimensions are calculated to construct a real-time updated multi-dimensional cooling state vector; The real-time cooling state vector is compared with a preset ideal cooling vector, deviation indexes of each dimension are calculated, and the cooling state is determined to be abnormal or performance to decline based on the deviation indexes; If it is determined that an abnormality occurs, the control instructions are automatically generated by combining the liquid cooling, gas cooling and air cooling device models in the cooling device calibration module, so as to dynamically adjust the liquid flow, gas pressure, air speed and wind direction distribution, so as to optimize the cooling stability, thermal distribution uniformity, thermal conductivity coefficient and profile deformation amount; Based on the historical cooling data and experimental verification results, the weight distribution of each dimension of the multi-dimensional cooling state vector is adjusted in real time to improve the function matching accuracy and cooling effect; When the cooling state continuously deviates from the preset threshold, an alarm signal is generated and a preset emergency cooling function is triggered, and abnormal data is recorded to a log database for subsequent analysis and function library optimization module to update the cooling function parameters.

[0010] The beneficial effects of the present application are: 1. The present application obtains detailed physical parameters of the profile and cooling targets through the parameter acquisition module, and the system can dynamically adjust the output of the cooling device according to the real-time collected cooling data, so as to ensure that the cooling process always follows the preset ideal path, effectively avoiding the defects of overcooling, undercooling and uneven cooling caused by mismatching of cooling intensity in traditional methods, and significantly improving the mechanical properties, dimensional accuracy and surface quality of the product; 2. By establishing a high-precision cooling device control model, the control lag and overshoot problem is solved. Through the special cooling device calibration module, the output characteristics of various cooling devices such as liquid cooling, fog cooling and air cooling are accurately tested and modeled, and an accurate corresponding relationship model between the control instructions and the actual output parameters is established. This makes the instructions issued by the data processing and control module able to be accurately predicted and executed, greatly reducing the adjustment lag and overshoot phenomenon caused by the nonlinear response of the device, improving the real-time performance and stability of the control, and providing a solid foundation for realizing high-quality cooling.

[0011] 3. The introduction of an intelligent, multi-dimensional state perception and decision-making mechanism enhances the system's adaptability. Rather than relying solely on a single temperature metric, the system constructs a cooling state vector that integrates multiple dimensions, including cooling stability, heat distribution uniformity, and thermal conductivity. This vector is then intelligently matched with a pre-set function library. This approach provides a more comprehensive and insightful depiction of the cooling process's true state, enabling the system to make optimal decisions for complex, nonlinear cooling processes with judgment capabilities similar to expert experience. Its adaptability far exceeds that of existing systems, which are limited to regional quantitative control.

[0012] 4. A complete closed-loop feedback and continuous optimization ecosystem has been formed, ensuring the long-term reliability and advancement of the system. The present invention not only includes online feedback correction, but also designs an offline function library optimization module. By analyzing historical data, the system can self-evaluate the effects of each control function and continuously optimize the vector-function mapping relationship using machine learning technology. This enables the system to continuously accumulate production experience, becoming more intelligent with use, and ultimately achieving the transition from automatic control to autonomous optimization, solving the pain point of traditional systems that become rigid once set and cannot adapt to new processes or aging equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 Schematic diagram of the overall architecture of the extrusion cooling online detection and feedback correction system.

[0014] Figure 2 This is a structural diagram of the specific process of the cooling device calibration module.

[0015] Figure 3 Schematic diagram of the flow of data processing and control module.

[0016] Figure 4 Schematic diagram of feedback optimization and fault diagnosis process.

[0017] Figure 5 This is a flowchart of the function library optimization mechanism. DETAILED DESCRIPTION

[0018] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0019] It should be understood that when an element is referred to as being "on" another element, it can be directly on the other element or intervening elements can also be present. In addition, it should be understood that when an element is referred to as being "connected, coupled, or "disposed" to another element, it can be directly connected, coupled, or disposed to the other element or intervening elements can also be present. As used herein, the terms "vertical," "horizontal," "left," "right," and the like are merely for purposes of illustration.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in the description herein is for describing particular embodiments only and is not intended to be limiting of the application. As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items.

[0021] The disclosed extrusion cooling online detection and feedback correction system, please see Figures 1 to 5 , through the coordination of multiple functional modules, realizes the whole cycle monitoring, analysis and control of the metal profile extrusion cooling process, and the overall structure includes parameter acquisition module, cooling data acquisition module, cooling device calibration module, data processing and control module, cooling device controller, feedback and optimization module, function library optimization module and cooling dynamic optimization module. The functions, key term definitions, technical details and mutual coordination of each module are described as follows: The core working logic of the system is: first, the inherent physical properties of the metal profile to be cooled and the cooling target requirements are obtained through the parameter acquisition module, providing a basis for subsequent cooling control strategy formulation; then the cooling data acquisition module collects key data in real time during the cooling process, reflecting the current cooling state in real time; the cooling device calibration module tests and establishes the quantitative relationship between the control instructions and the actual output parameters of the cooling device in advance, ensuring the accuracy of the control instructions; the data processing and control module analyzes and calculates the collected cooling data, constructs a multi-dimensional cooling state vector that can comprehensively represent the cooling state, and then matches the vector with the execution functions in the pre-set cooling control function library to generate control instructions that adapt to the current cooling state; the cooling device controller receives the control instructions and drives the cooling device to adjust the operating parameters, realizing closed-loop control of the cooling process; at the same time, the feedback and optimization module continuously monitors the cooling effect and makes parameter adjustments or function re-matching, and the fault diagnosis unit simultaneously monitors the device operating state and handles exceptions; the function library optimization module optimizes the cooling control function library based on historical data, and the machine learning model further improves the matching accuracy of the vector and the function; the cooling dynamic optimization module adjusts the vector weight and cooling parameters in real time to ensure that the cooling process is always in the optimal state.

[0022] The core function of the parameter acquisition module is to obtain the physical parameters of the metal profile to be cooled, including material thermal conductivity, specific heat capacity, cross-sectional structure, initial temperature, target cooling temperature, and cooling time requirement. The material thermal conductivity is an inherent physical quantity that characterizes the heat conduction ability of the metal profile, which refers to the heat passing through a unit area per unit time under a unit temperature gradient. The specific heat capacity is a physical quantity that characterizes the heat required to raise or lower the temperature of a unit mass of metal profile by 1 degree Celsius, which is also a material-specific property. The cross-sectional structure refers to the cross-sectional shape and corresponding size parameters of the metal profile to be cooled. The cross-sectional shape includes rectangular, circular, I-shaped, and channel-shaped structures, and the size parameters need to be determined according to the shape. For example, the size parameters of a rectangular cross-section include the length, width, and thickness, while the size parameters of a circular cross-section include the diameter, and the size parameters of an I-shaped cross-section include the web height, web thickness, flange width, and flange thickness.

[0023] The initial temperature refers to the average surface and internal temperature of the metal profile immediately after it is extruded from the extruder and enters the cooling stage. For example, the initial temperature of an aluminum alloy profile after extrusion is generally between 450 and 550 degrees Celsius. Non-contact or contact temperature measurement equipment is used to obtain the initial temperature. Considering the high surface temperature of the profile after extrusion and the need to quickly enter the cooling stage, an infrared thermometer is preferred for multi-point temperature measurement on the profile surface. The number of temperature measurement points is determined according to the cross-sectional structure of the profile. For example, a rectangular cross-section can have six temperature measurement points at the center, edges, and four corner points. The average temperature of multiple points is calculated as the initial temperature. The measurement range of the infrared thermometer should cover the initial temperature range of the profile, usually 0 to 1000 degrees Celsius to meet the requirements.

[0024] The target cooling temperature refers to the average surface and internal temperature of the metal profile after the cooling process is completed. This temperature needs to be set according to the requirements of subsequent processing technology. For example, if an aluminum alloy profile needs to be stretched, the target cooling temperature is usually set between 80 and 120 degrees Celsius. The target cooling temperature is determined according to the subsequent processing technology file, such as the temperature requirements of each process marked in the profile processing process provided by the production planning department. The parameter acquisition module obtains this parameter by reading the process file data in the production management system.

[0025] The cooling duration requirement is a parameter used to define the time constraint of the cooling process in the present application, which includes the total time of the cooling process and the time distribution requirement of the cooling rate, wherein the total time of the cooling process refers to the total length of time required for the profile to enter the cooling link to the temperature of the profile to drop to the target cooling temperature; the time distribution requirement of the cooling rate refers to the temperature drop rate requirement corresponding to different time periods within the total cooling time, which is set to avoid excessive thermal stress in the profile due to too fast cooling rate, thereby causing deformation or cracking, while avoiding the impact on production efficiency due to too slow cooling rate, for example, for a 6061 aluminum alloy rectangular profile, the time distribution requirement of the cooling rate can be set to a cooling rate of 2 degrees Celsius per second within the first 100 seconds when the temperature drops from 500 degrees Celsius to 300 degrees Celsius, a cooling rate of 1.5 degrees Celsius per second within the middle 100 seconds when the temperature drops from 300 degrees Celsius to 150 degrees Celsius, and a cooling rate of 0.5 degrees Celsius per second within the last 100 seconds when the temperature drops from 150 degrees Celsius to 100 degrees Celsius. The cooling duration requirement is also determined according to the production process file and the profile characteristics, and the process personnel set the cooling duration requirement according to the thickness of the profile, the greater the thickness, the slower the heat dissipation, the longer the total cooling time, the greater the thermal conductivity of the material, the faster the heat dissipation, the shorter the total cooling time, and the beat requirement of subsequent processing, and store the cooling duration requirement in the system database, and the parameter acquisition module reads the parameter from the database.

[0026] The core function of the cooling data acquisition module is to acquire cooling data in real time during the cooling process, which is the key basis for reflecting the current cooling state and calculating the cooling stability and uniformity. The cooling data includes temperature fluctuation data, thermal distribution data and heat conduction data. The temperature fluctuation data refers to the fluctuation of the temperature of the specified monitoring point of the metal profile to be cooled with time during the cooling process, which is used to judge the stability of the cooling process. The temperature fluctuation data acquisition needs to set multiple fixed monitoring points on the profile to be cooled. The number and position of the monitoring points are determined according to the cross-sectional structure of the profile. If the temperature fluctuation amplitude of the same monitoring point in the continuous time period is small, it means that the cooling state is stable, and if the fluctuation amplitude is large, it means that there is an abnormality in the cooling process, such as unstable cooling medium flow, cooling device failure, etc.

[0027] Thermal distribution data refers to the temperature distribution at different locations on the cross-section or surface of the metal profile being cooled at a certain moment during the cooling process. Thermal distribution data is collected using an infrared thermal imager. The camera is mounted on the observation window of the cooling device, with the lens angle adjusted to focus on the cross-section of the profile being cooled. The acquisition frequency is set based on the dynamic changes in the cooling process, typically 0.5 Hz, meaning a thermal image is captured every 2 seconds. The infrared thermal imager converts the captured thermal images into temperature distribution data, which contains the temperature value corresponding to each pixel on the profile's cross-section. For example, if a pixel has a temperature of 450°C, its adjacent pixel has a temperature of 448°C. The collection of these temperature values ​​is the thermal distribution data. This data allows for intuitive observation of temperature differences between the center and edges of the profile, as well as between different locations. This data is used to determine cooling uniformity. Small temperature differences between different locations indicate uniform heat distribution and good cooling. Large differences indicate localized excessive or insufficient cooling, which may lead to profile deformation. Heat conduction data refers to the rate at which heat is transferred within the metal profile and between the profile and the cooling medium, such as coolant or cooling gas, during the cooling process. This data is used to reflect the efficiency of heat transfer during the cooling process. The greater the heat conduction rate, the faster the heat dissipation and the higher the cooling efficiency. The heat conduction data is collected using a heat flux meter, and a heat flux meter with a range of 0 to 1000 watts per square meter and an accuracy of ±5% is selected to meet the measurement requirements; the sensing surface of the heat flux meter is attached to the surface of the profile to be cooled, and the attachment position is adjacent to the thermocouple monitoring point to facilitate the correlation of temperature data and heat flow data; the data acquisition module is connected to the heat flux meter through a data cable, and the acquisition frequency is set. The acquisition frequency is consistent with the temperature fluctuation data at 1 Hz, and the heat flux density value is obtained in real time. The heat flux density value is the heat conduction data at the current moment, reflecting the rate at which heat is transferred from the profile surface to the cooling medium at this time; at the same time, combined with the temperature difference in the temperature fluctuation data, that is, the difference between the profile surface temperature and the cooling medium temperature, the accuracy of the heat conduction data can be further verified. The verification is based on Fourier's law, which states that the heat flux density is equal to the thermal conductivity multiplied by the temperature difference divided by the heat transfer distance, where the heat transfer distance is the distance from the profile surface to the inside, which can be determined by the cross-sectional size.

[0028] The core function of the cooling device calibration module is to test the output characteristics of the liquid cooling device, the fog cooling device and the air cooling device, and to establish a corresponding relationship model between the control instructions and the actual output parameters of the cooling device. The role of the module is to ensure that the control instructions generated by the subsequent data processing and control module can be accurately converted into the actual output of the cooling device, avoiding the decline of control accuracy due to the deviation of device characteristics. The liquid cooling device, the fog cooling device and the air cooling device are execution devices for realizing profile cooling. The liquid cooling device uses liquid as the cooling medium, including multiple frequency conversion water pumps, intelligent positioning proportional water valves and pressure sensors, for providing cooling water and adjusting water speed; the fog cooling device atomizes the cooling liquid into water mist through a nozzle and sprays it on the surface of the profile to be cooled; the air cooling device uses environmental air as the cooling medium, and its components include fans such as centrifugal fans, axial flow fans, wind deflectors and wind speed adjusting mechanisms. The working principle is that the fan generates airflow after being powered on, and the airflow is guided by the wind deflector to blow at the profile surface at a preset wind speed and direction, achieving heat dissipation through heat exchange between air and the profile.

[0029] The output characteristics refer to the change rule of the actual output parameters of the cooling device under the action of different control instructions, such as the cooling liquid flow of the liquid cooling device, the airflow speed of the fog cooling device and the wind speed of the air cooling device. The corresponding relationship model is a model that describes the quantitative relationship between the control instructions and the actual output parameters of the cooling device using mathematical formulas or curves. This model is a bridge connecting the control instructions and the actual action of the device. For example, the control instruction of the liquid cooling device is the control current of the delivery pump, and the actual output parameter is the cooling liquid flow. The corresponding relationship model can be a linear formula of flow and control current. Through this formula, the corresponding control current can be calculated according to the required flow, or the actual flow can be predicted according to the control current.

[0030] The calibration process of the cooling device calibration module needs to be completed before the system is formally operated, and needs to be repeated once every 3 months to compensate for the changes in output characteristics caused by device aging such as wear of the delivery pump and dust accumulation on the fan blades. The above-mentioned correction relationship model of the liquid cooling device, the mapping relationship model of the fog cooling device and the correlation model of the air cooling device are stored in the calibration database, and the storage content includes the mathematical expression of the model, the specific value of each coefficient and the calibration conditions corresponding to each model such as calibration time, environmental temperature and cooling medium type, which facilitates the reference when the model is called later, and ensures the accuracy of the application range of the model.

[0031] The data processing and control module is the core control unit of the system, and its core function is to receive the cooling data collected by the cooling data acquisition module, analyze and calculate the cooling stability, heat distribution uniformity and heat transfer coefficient, construct a multi-dimensional cooling state vector, and then match the vector with the execution functions in the preset cooling control function library to generate control instructions.

[0032] The cooling stability refers to the ability of the temperature of the metal profile to be cooled to remain stable and have a small fluctuation range during the cooling process. The index is quantified by a temperature fluctuation coefficient. The temperature fluctuation coefficient is the ratio of the temperature standard deviation to the average temperature, and the calculation formula is temperature fluctuation coefficient equals temperature standard deviation divided by average temperature. The temperature fluctuation coefficient is a dimensionless parameter. The temperature standard deviation is the square root of the average of the square sum of the deviation between a plurality of temperature values collected in a certain time period and the average of the temperature values. The unit is consistent with the temperature unit, which is degrees Celsius. The average temperature is the arithmetic mean of a plurality of temperature values collected in the time period, and the unit is also degrees Celsius. The calculation process of the temperature standard deviation is as follows: first, a plurality of temperature values in the temperature fluctuation data in a certain time period, such as 10 seconds, are selected. The number of selected temperature values is not less than 5 to ensure statistical effectiveness. The average of the temperature values, i.e., the average temperature, is calculated. The average temperature equals the sum of all temperature values divided by the number of temperature values. Then, the deviation of each temperature value from the average is calculated. The deviation equals the single temperature value minus the average temperature. Next, the square sum of all deviations is calculated. The square sum equals the sum of the squares of each deviation. Then, the variance is calculated. If the selected temperature values are sample data, the variance equals the square sum divided by the number of temperature values minus 1. If they are total data, the variance equals the square sum divided by the number of temperature values. In practical applications, the sample variance is usually used. Finally, the temperature standard deviation equals the square root of the variance. In order to clearly understand the calculation process, an operation example is given. Assuming that the temperature data of a monitoring point collected in 10 seconds is 200 degrees Celsius, 202 degrees Celsius, 198 degrees Celsius, 201 degrees Celsius, and 199 degrees Celsius, a total of 5 temperature values, first, the average temperature is calculated. The average temperature equals the sum of 200, 202, 198, 201, and 199 divided by 5, and the calculation result is 200 degrees Celsius. Then, the deviation of each temperature value from the average is calculated. The deviations are 200 minus 200 equal to 0 degrees Celsius, 202 minus 200 equal to 2 degrees Celsius, 198 minus 200 equal to negative 2 degrees Celsius, 201 minus 200 equal to 1 degree Celsius, and 199 minus 200 equal to negative 1 degree Celsius. Next, the square sum of the deviations is calculated. The square sum of 0, 2, negative 2, 1, and negative 1 is 0 plus 4 plus 4 plus 1 plus 1 equal to 10 degrees Celsius squared. Then, the sample variance is calculated. The sample variance equals 10 divided by 5 minus 1, which is 10 divided by 4, equal to 2.5 degrees Celsius squared. Finally, the temperature standard deviation is calculated. The temperature standard deviation equals the square root of 2.5, which is about 1.58 degrees Celsius. The temperature fluctuation coefficient equals 1.58 divided by 200, which is about 0.0079, i.e., 0.79%. The smaller the value, the smaller the temperature fluctuation and the better the cooling stability. Generally, the temperature fluctuation coefficient is required to be not greater than 0.01, i.e., 1%. If the temperature fluctuation coefficient is greater than 0.01, the cooling parameters need to be adjusted to improve the stability.

[0033] The heat distribution uniformity refers to the uniformity of temperature distribution at different positions of the cross section or surface of the metal profile to be cooled during the cooling process. The heat distribution uniformity is quantified by the temperature distribution uniformity, and the calculation formula of the temperature distribution uniformity is equal to 1 minus the quotient of the temperature sample variance divided by the temperature population variance. The temperature distribution uniformity is a dimensionless parameter and the value range is 0 to 1. The closer the value is to 1, the more uniform the heat distribution is. The temperature sample variance refers to the sample variance of the temperature values of a plurality of sampling points on the cross section of the heat distribution data at a certain moment. The number of sampling points is not less than 10 to ensure covering the entire cross section. The calculation method is consistent with the sample variance in the above temperature standard deviation. The temperature population variance refers to the preset value of the temperature population variance of the profile to be cooled in the current cooling stage. The preset value is determined according to the material quality and cooling quality requirements of the profile. For example, the temperature population variance preset value of the aluminum alloy profile can be set to 4 degrees Celsius squared, corresponding to a population standard deviation of 2 degrees Celsius. In order to clearly understand the calculation process, an operation example is given: the temperature values of 10 sampling points on the cross section of the heat distribution data at a certain moment are 150 degrees Celsius, 151 degrees Celsius, 149 degrees Celsius, 152 degrees Celsius, 148 degrees Celsius, 150 degrees Celsius, 151 degrees Celsius, 149 degrees Celsius, 150 degrees Celsius and 151 degrees Celsius. First, calculate the sample mean of these temperature values, which is equal to 150 plus 151 plus 149 plus 152 plus 148 plus 150 plus 151 plus 149 plus 150 plus 151 divided by 10, and the result is 150 degrees Celsius. Then calculate the sample variance, which is equal to the sum of the squares of the deviations of each temperature value from the sample mean divided by 10 minus 1. The sum of the squares of the deviations is 0 squared plus 1 squared plus negative 1 squared plus 2 squared plus negative 2 squared plus 0 squared plus 1 squared plus negative 1 squared plus 0 squared plus 1 squared, which is equal to 0 plus 1 plus 1 plus 4 plus 4 plus 0 plus 1 plus 1 plus 0 plus 1, which is equal to 13 degrees Celsius squared. The sample variance is equal to 13 divided by 9, which is approximately 1.44 degrees Celsius squared. Given that the temperature population variance is 4 degrees Celsius squared, the heat distribution uniformity is equal to 1 minus the quotient of 1.44 divided by 4, which is 1 minus 0.36, equal to 0.64, which is 64%. If the heat distribution uniformity is required to be not less than 0.8, i.e. 80%, the current heat distribution uniformity does not meet the requirements, and the output parameters of the cooling device need to be adjusted, such as adjusting the position of the liquid cooling nozzle to uniformly spray the cooling liquid on the surface of the profile to improve the uniformity.

[0034] The heat transfer coefficient is a parameter reflecting the efficiency of heat transfer in the metal profile and between the profile and the cooling medium during the cooling process. The heat transfer coefficient is calculated from the heat transfer data collected by the cooling data collection module, i.e. heat flux and temperature data. The calculation formula is heat transfer coefficient equals heat flux divided by the ratio of temperature difference and heat transfer distance. The unit of heat transfer coefficient is watt per meter kelvin. Heat flux is the heat per unit area per unit time, and its unit is watt per square meter, which is collected by a heat flow meter. The temperature difference is the difference between the surface temperature of the profile and the temperature of the cooling medium, and its unit is kelvin. The surface temperature of the profile is collected by a thermocouple, and the temperature of the cooling medium is collected by a temperature sensor installed in the cooling device. The heat transfer distance is the length of the heat transfer path, and its unit is meter. For the heat transfer between the profile surface and the cooling medium, the heat transfer distance is the distance from the profile surface to the cooling medium, such as the spraying distance of the cooling liquid, which is usually 0.01 meters. For the heat transfer inside the profile, the heat transfer distance is half the thickness of the profile section, such as half the thickness of a rectangular section. The theoretical basis of this formula is Fourier's law, which states that heat flux is equal to the negative thermal conductivity coefficient multiplied by the temperature gradient, which is the ratio of temperature difference to heat transfer distance. Since the heat flow direction is from high temperature to low temperature, the absolute value is taken here. To clearly understand the calculation process, an operation example is given: the heat flux collected by the heat flow meter is 800 watts per square meter, the surface temperature of the profile collected by the thermocouple is 200 degrees Celsius, i.e. 473.15 kelvin, and the temperature of the cooling medium, i.e. the cooling liquid, is 25 degrees Celsius, i.e. 298.15 kelvin. Then the temperature difference is equal to 473.15 kelvin minus 298.15 kelvin, which is 175 kelvin. The heat transfer distance is 0.01 meters, i.e. the spraying distance of the cooling liquid in liquid cooling. Then the heat transfer coefficient is equal to 800 divided by the ratio of 175 to 0.01, i.e. 800 divided by 17500, which is about 0.0457 watts per meter kelvin. It should be noted that the heat transfer coefficient here is an equivalent coefficient considering the thermal conductivity of the profile material, the thermal conductivity of the cooling medium and the contact thermal resistance, which is different from the inherent thermal conductivity of the material. Its value will change with factors such as the type of cooling medium and flow rate, and is used to reflect the heat transfer efficiency of the current cooling process.

[0035] The multi-dimensional cooling state vector is a mathematical vector used to comprehensively represent the state of the cooling process at a certain moment, the dimension of the vector is composed of key indicators that can reflect the cooling state, according to the definition of the claim, the initial core dimension of the multi-dimensional cooling state vector includes cooling stability, that is, represented by temperature fluctuation coefficient, thermal distribution uniformity, that is, represented by temperature distribution uniformity, and thermal conductivity coefficient, so the initial vector is a three-dimensional vector, denoted as the multi-dimensional cooling state vector is equal to the vector composed of temperature fluctuation coefficient, thermal distribution uniformity and thermal conductivity coefficient; The process of constructing the multi-dimensional cooling state vector is to first determine the update period of the vector, the update period matches the collection frequency of the cooling data, for example, if the cooling data collection frequency is 1 hertz, the vector update period is 1 second, that is, a new multi-dimensional cooling state vector is constructed every 1 second; Then in each update period, the latest temperature fluctuation data, thermal distribution data and thermal conductivity data are obtained from the cooling data collection module; Then the temperature fluctuation coefficient, the thermal distribution uniformity and the thermal conductivity coefficient are calculated respectively according to the above calculation method; Finally, the three parameters are combined as the three components of the vector to form the multi-dimensional cooling state vector; For example, at a certain moment, the temperature fluctuation coefficient is 0.0079, the thermal distribution uniformity is 0.64, and the thermal conductivity coefficient is 0.0457 watt per meter kelvin, then the multi-dimensional cooling state vector at this moment is a vector composed of 0.0079, 0.64 and 0.0457; Subsequently, with the expansion of the system function, the multi-dimensional cooling state vector will add new dimensions such as profile deformation amount dimension, and the extended vector dimension will be increased accordingly.

[0036] The cooling control function library is a collection of a plurality of execution functions, each execution function corresponding to a cooling control strategy, that is, the execution function can output a set of adaptive cooling device control parameters such as the control current of the liquid cooling device, the gas pressure set value of the fog cooling device, and the fan speed of the air cooling device when a given multi-dimensional cooling state vector is input, and these control parameters are determined based on a large amount of experimental data and process experience, and can adjust the cooling process to the ideal state; the construction process of the cooling control function library is as follows: first, determine the input and output of the execution function, the input of the execution function is a multi-dimensional cooling state vector, and the output is a liquid cooling control parameter, that is, the control current of the liquid cooling device, a gas cooling control parameter, that is, the gas pressure set value of the fog cooling device, and an air cooling control parameter, that is, the fan speed of the air cooling device, that is, the execution function can be expressed as the result of the execution function acting on the multi-dimensional cooling state vector being equal to the vector composed of the liquid cooling control parameter, the gas cooling control parameter, and the air cooling control parameter; then, the optimal control parameters corresponding to different multi-dimensional cooling state vectors are obtained through experiments. The orthogonal experimental design is used in the experiment, which is an efficient experimental design method that can cover different level combinations of multiple factors through a small number of experiments. The value range of each dimension of the multi-dimensional cooling state vector is selected, such as the temperature fluctuation coefficient being 0.005 to 0.02, the thermal distribution uniformity being 0.5 to 0.95, and the thermal conductivity coefficient being 0.03 to 0.06 watts per meter kelvin. A number of levels are selected within the range to form a plurality of experimental conditions, such as the temperature fluctuation coefficient taking 3 levels of 0.005, 0.01, and 0.02, the thermal distribution uniformity taking 3 levels of 0.5, 0.75, and 0.95, and the thermal conductivity coefficient taking 3 levels of 0.03, 0.045, and 0.06, a total of 3*3*3 equaling 27 experimental conditions. For each experimental condition, adjust the control parameters of the liquid cooling, gas cooling, and air cooling devices, and monitor the cooling effect, such as whether the multi-dimensional cooling state vector at the subsequent time is closer to the ideal cooling vector. Through multiple adjustments, the control parameter combination that optimizes the cooling effect is found, and the control parameter combination is taken as the output of the execution function corresponding to the multi-dimensional cooling state vector. For example, when the experimental condition is that the multi-dimensional cooling state vector is equal to the vector composed of 0.0079, 0.64, and 0.0457, the optimal control parameter combination is 3 amperes of liquid cooling control current, 0.5 megapascals of gas cooling gas pressure set value, and 2000 revolutions per minute of air cooling fan speed. Therefore, the result of the corresponding execution function acting on the vector is the vector composed of 3 amperes, 0.5 megapascals, and 2000 revolutions per minute. Store all the execution functions corresponding to the experimental conditions in the cooling control function library, and set the corresponding vector interval for each execution function, that is, the range of the multi-dimensional cooling state vector applicable to the execution function, such as the above execution function corresponding to the vector interval of the temperature fluctuation coefficient being 0.006 to 0.009, the thermal distribution uniformity being 0.6 to 0.68, and the thermal conductivity coefficient being 0.04 to 0.05 watts per meter kelvin, to ensure that candidate execution functions can be quickly screened out during subsequent matching.

[0037] The data processing and control module comprises a data preprocessing unit, a vector calculation unit and a function matching unit. The function of the data preprocessing unit is to filter and remove outliers from the cooling data. The filtering process uses the mean filtering method. Mean filtering is a commonly used linear filtering method. The mean value of all data in a certain size sliding window is taken as the filtering result of the center data of the window. The size of the sliding window is determined according to the noise of the cooling data. Usually, 5 is selected, that is, the window contains the current data and the previous and next 2 data. For example, for the temperature fluctuation data sequence 200, 202, 198, 201, 199, 203, if the window size is 5, the first window data is 200, 202, 198, 201, 199, and the average value is 200. The filtered first center data 198 becomes 200. The outlier removal process uses the 3σ criterion. The 3σ criterion is an outlier detection method based on normal distribution. First, the mean and standard deviation of the data sequence are calculated. Then, the data outside the range of the mean plus or minus 3 times the standard deviation is determined as an outlier and is removed. The missing gap after removal is filled with the average value of the adjacent data. For example, the average value of a certain temperature data sequence is 200 degrees Celsius, the standard deviation is 1.58 degrees Celsius, and 3 times the standard deviation is 4.74 degrees Celsius. The data outside the range of 200 minus 4.74 equal to 195.26 degrees Celsius to 200 plus 4.74 equal to 204.74 degrees Celsius is determined as an outlier. The function of the vector calculation unit is to calculate the index value of each dimension of the multi-dimensional cooling state vector based on the preset quantization formula. That is, according to the calculation methods of the cooling stability, the thermal distribution uniformity and the thermal conductivity coefficient, the preprocessed cooling data is substituted into the formula to calculate the numerical value of each dimension, and then the multi-dimensional cooling state vector is formed. The function of the function matching unit is to match the multi-dimensional cooling state vector with the execution function and generate a control instruction. The specific working process is as follows. First, read the latest multi-dimensional cooling state vector, that is, the latest vector, and read the multi-dimensional cooling state vector of the last period, that is, the last period vector. Then, calculate the Euclidean distance between the latest vector and the last period vector. The Euclidean distance is an index for measuring the difference between two vectors in space. The greater the difference, the greater the change in the cooling state, which requires a new match of the execution function. The calculation formula of the Euclidean distance is the square root of the sum of the squares of the difference between the latest value and the last period value of each dimension. Since the temperature fluctuation coefficient and the thermal distribution uniformity are dimensionless parameters, and the thermal conductivity coefficient has a unit, the thermal conductivity coefficient needs to be normalized before calculation. The normalization method is to divide the thermal conductivity coefficient by the maximum value of its value range to convert it into a dimensionless value between 0 and 1. For example, the maximum value of the thermal conductivity coefficient is 0.06 watt per meter kelvin. If a certain thermal conductivity coefficient is 0.0457 watt per meter kelvin, the normalized value is 0.0457 divided by 0.06, which is about 0.762. Then, a preset threshold value of the Euclidean distance is set. The threshold value is determined according to the stability requirement of the cooling process. For example, it is set to 0.02, if the calculated Euclidean distance exceeds the preset threshold or the latest vector exceeds the vector interval corresponding to the current execution function, the matching process is started, if it does not exceed the threshold and is within the interval, it is not necessary to re-match, continue to use the current execution function; after starting the matching process, the matching strategy based on the preset vector interval is used to screen the candidate execution function, the screening rule is to traverse all execution functions in the cooling control function library, judge whether the latest vector falls within the vector interval corresponding to the execution function, if it falls within the interval, the execution function is used as the candidate execution function; then calculate the cosine similarity between the latest vector and the preset vector of each candidate execution function, the preset vector of the candidate execution function is the ideal multi-dimensional cooling state vector corresponding to the function, the cosine similarity is an index for measuring the consistency of two vectors, the higher the consistency, the closer the cosine similarity to 1, which means that the execution function is more suitable for the current cooling state, the formula for calculating the cosine similarity is that the cosine similarity is equal to the sum of the product of each dimension of two vectors divided by the product of the modulus of two vectors, wherein the vector modulus is equal to the square root of the sum of the square of each dimension value, when calculating, the thermal conductivity coefficient also needs to be normalized; finally, select the execution function with the highest cosine similarity as the target function, and combine the corresponding relationship model in the cooling device calibration module to convert the output parameters of the target function into control instructions, for example, the output parameters of the target function are liquid cooling control current 3 amperes, according to the correction relationship model of the liquid cooling device, the flow rate is equal to 10 times the control current, so 3 amperes corresponds to the flow rate of 30 liters per minute, therefore, the control parameter of the liquid cooling device in the control instruction is the control current of 3 amperes corresponding to the flow rate of 30 liters per minute, and the control parameters of the gas cooling and air cooling devices are converted in the same way, and finally integrated to form a complete control instruction.

[0038] The core function of the cooling device controller is to receive the control instructions generated by the data processing and control module, convert the control instructions into control signals that can be recognized and executed by the cooling device, drive the liquid cooling device, the fog cooling device and the air cooling device to adjust the operating parameters, and ensure that the cooling device works according to the requirements of the control instructions. The cooling device controller uses an industrial programmable logic controller (PLC) as the core control unit. The industrial PLC has the characteristics of strong anti-interference ability, high stability, and adaptation to harsh industrial production environments such as high temperature and dust. The controller also includes a signal input module, a signal output module, a communication module, and a human-machine interaction module. The signal input module is used to receive the control instructions transmitted by the data processing and control module, usually through Ethernet communication. The signal output module is used to output control signals to the actuators of the cooling device, such as the delivery pump, air compressor, and fan. The communication module is used to realize two-way communication with the data processing and control module and the calibration database, ensuring the reception of control instructions and the uploading of operating state data. The human-machine interaction module includes a touch screen, which is used to display the current operating parameters of the cooling device, such as the cooling liquid flow, air pressure, wind speed, and the execution status of the control instructions. It also supports manual input of parameters or modification of set values by the operator, facilitating monitoring and emergency operation.

[0039] The control instruction received by the cooling device controller includes the control current of the liquid cooling device, the gas pressure set value of the fog cooling device, and the fan speed of the air cooling device. The controller needs to convert these parameters into corresponding control signals. For the liquid cooling device, the control current in the control instruction is an analog parameter. The controller outputs an analog current signal of 4 to 20 milliamperes to the frequency converter of the delivery pump through the signal output module. The frequency converter adjusts the output current according to the size of the analog signal to control the speed of the delivery pump, thereby adjusting the flow of the cooling liquid. The corresponding relationship between the analog signal and the control current is determined according to the control current range. For example, when the control current range is 0 to 5 amperes, 4 milliamperes correspond to 0 amperes of control current, and 20 milliamperes correspond to 5 amperes of control current. Therefore, if the control current in the control instruction is 3 amperes, the corresponding analog signal is 4 milliamperes plus 3 amperes divided by 5 amperes multiplied by 20 milliamperes minus 4 milliamperes, which is 4 plus 3 divided by 5 multiplied by 16, equaling 13.6 milliamperes. At the same time, the controller collects the actual cooling liquid flow through the electromagnetic flowmeter and compares it with the target flow in the control instruction calculated according to the calibration model. If there is a deviation, such as an actual flow of 28 liters per minute and a target flow of 30 liters per minute, a deviation of 2 liters per minute, the output analog signal is adjusted through the PID adjustment algorithm. The PID adjustment algorithm is a commonly used closed-loop control algorithm that calculates and adjusts the amount through the combination of proportional, integral, and differential links to make the actual flow approach the target flow, ensuring that the flow control accuracy reaches ±0.5 liters per minute. For the fog cooling device, the gas pressure set value in the control instruction is an analog parameter. The controller outputs an analog current signal of 4 to 20 milliamperes to the electrical positioner of the gas pressure regulating valve. The electrical positioner adjusts the opening of the regulating valve according to the signal size to control the gas pressure. The corresponding relationship between the analog signal and the gas pressure set value is determined according to the gas pressure range. For example, when the gas pressure range is 0.2 to 1.0 megapascals, 4 milliamperes correspond to 0.2 megapascals, and 20 milliamperes correspond to 1.0 megapascals. Therefore, if the gas pressure set value in the control instruction is 0.5 megapascals, the corresponding analog signal is 4 milliamperes plus 0.5 megapascals minus 0.2 megapascals divided by 1.0 megapascals minus 0.2 megapascals multiplied by 16 milliamperes, which is 4 plus 0.3 divided by 0.8 multiplied by 16, equaling 10 milliamperes. At the same time, the controller collects the actual gas pressure through the pressure sensor and compares it with the target gas pressure. The PID adjustment algorithm is used to adjust the opening of the regulating valve to ensure that the gas pressure control accuracy reaches ±0.01 megapascals.For the air cooling device, the fan speed in the control instruction is an analog parameter, and the controller outputs an analog current signal of 4 to 20 milliamps to the frequency converter of the fan. The frequency converter adjusts the fan speed according to the signal. The corresponding relationship between the analog signal and the fan speed is determined according to the speed range. For example, when the speed range is 1000 to 3000 revolutions per minute, 4 milliamps corresponds to 1000 revolutions per minute, and 20 milliamps corresponds to 3000 revolutions per minute. Therefore, if the fan speed in the control instruction is 2000 revolutions per minute, the corresponding analog signal is 4 milliamps plus 2000 revolutions per minute minus 1000 revolutions per minute divided by 3000 revolutions per minute minus 1000 revolutions per minute multiplied by 16 milliamps, i.e. 4 plus 1000 divided by 2000 multiplied by 16 equals 12 milliamps. At the same time, the controller collects the actual fan speed through the speed sensor and compares it with the target speed. The frequency converter output is adjusted using the PID adjustment algorithm to ensure that the speed control accuracy reaches ±10 revolutions per minute.

[0040] The cooling device controller monitors the running state of the cooling device in real time. Running parameters such as the delivery pump current of the liquid cooling device, the air compressor pressure of the fog cooling device, and the fan temperature of the air cooling device are collected through sensors installed on each device. The data collected by the sensors is transmitted to the PLC through the signal input module. The controller judges the collected running parameters. If an abnormal state is detected, such as the delivery pump current of the liquid cooling device exceeding the rated value, the air pressure of the fog cooling device being lower than the lower limit value, or the fan temperature of the air cooling device exceeding the allowed range, a state signal is immediately generated and fed back to the data processing and control module and the feedback and optimization module through the communication module. At the same time, alarm information such as liquid cooling pump overload, air cooling air pressure too low, and air cooling fan over temperature is displayed on the touch screen. The alarm information includes the type of abnormality, the time of occurrence, and the current parameter value, which facilitates the operator to handle it in a timely manner. For example, if the actual air pressure of the fog cooling device is 0.25 megapascals, which is lower than the preset standard range of 0.3 to 0.7 megapascals, the controller generates a state signal indicating that the air cooling air pressure is low, which is fed back to the data processing and control module, triggering the subsequent compensation or fault diagnosis process.

[0041] The core function of the feedback and optimization module is to continuously monitor the cooling effect, adjust the parameters or match the function according to the difference between the cooling effect and the ideal state, and at the same time, the fault diagnosis unit handles the abnormal operation of the device. The working process of the feedback and optimization module is as follows: first, continuously collect the cooling data output by the cooling data collection module, and calculate the multi-dimensional cooling state vector in real time according to the calculation method of the data processing and control module; then compare the multi-dimensional cooling state vector calculated in real time with the ideal cooling vector, wherein the ideal cooling vector is a target state vector calculated according to the cooling target parameters, including the target cooling temperature, the target cooling rate, etc. The calculation basis of the ideal cooling vector is the heat transfer theory, for example, the ideal temperature at different times is calculated according to the Fourier heat equation, and then the ideal temperature fluctuation coefficient, the ideal heat distribution uniformity and the ideal heat conduction coefficient are determined according to the target cooling stability and uniformity requirements, and then combined to form the ideal cooling vector. The expression of the Fourier heat equation in one-dimensional case is ideal temperature equal to initial temperature minus heat conduction amount multiplied by time divided by density multiplied by specific heat capacity multiplied by volume. The ideal temperature is the basic data of the temperature-related dimension in the ideal cooling vector, the heat conduction amount is the heat transfer amount in the ideal state, the time is the cooling time, the density is the material density, the specific heat capacity is the material specific heat capacity, and the volume is the material volume; then calculate the Euclidean distance between the real-time multi-dimensional cooling state vector and the ideal cooling vector, and take the Euclidean distance as the cooling effect difference index. The calculation method of the Euclidean distance is consistent with that in the data processing and control module, and each dimension needs to be normalized; then different operations are performed according to the size of the difference index. If the difference index is within the preset allowable deviation threshold, the preset allowable deviation threshold is determined according to the cooling quality requirement, for example, 0.1, then the parameters of the current execution function are fine-tuned and optimized. The fine-tuned parameters include cooling time, liquid cooling flow coefficient, etc. The fine-tuning range is usually ±5% of the current parameter, for example, if the liquid cooling flow coefficient in the current execution function is 10, then the fine-tuned value is 9.5 to 10.5; if the difference index exceeds the preset allowable deviation threshold, trigger the data processing and control module to re-match the function, and ensure to obtain the execution function more suitable for the current cooling state; if the running parameters of the cooling device continuously deviate from the standard range, the duration is set according to the system stability requirement, for example, 30 seconds, then generate an alarm signal and call a preset emergency cooling function. The emergency cooling function is a pre-set control function with high cooling intensity, and its output parameters such as liquid flow, air pressure and wind speed are set to large values. At the same time, safety measures such as stopping the machine when the deformation of the profile exceeds the preset value are set to ensure that the profile is not damaged in an emergency.

[0042] The fault diagnosis unit is a subunit of the feedback and optimization module. The working process of the fault diagnosis unit is as follows: first, real-time monitoring of the operating parameters of the cooling device, the monitored parameters including the coolant flow of the liquid cooling device, the delivery pump current, the air pressure of the fog cooling device, the air compressor running state, the wind speed of the air cooling device, the fan speed and temperature, etc.; then comparing the monitored operating parameters with the preset standard range, the preset standard range being determined according to the rated parameters of the cooling device and the process requirements, for example, the standard range of the coolant flow of the liquid cooling device is 20 to 80 liters per minute, the standard range of the air pressure of the fog cooling device is 0.3 to 0.7 megapascals, and the standard range of the wind speed of the air cooling device is 5 to 12 meters per second; if a parameter deviates from the preset standard range, the corresponding model in the calibration database is called to calculate the compensation parameter, the calculation method of the compensation parameter being that the compensation amount is equal to the difference between the standard value and the actual value multiplied by the model coefficient, the model coefficient being the proportional coefficient in the calibration model, for example, the calibration model of the liquid cooling device is that the flow is equal to 10 times the control current, if the actual flow is 28 liters per minute and the standard flow is 30 liters per minute, the compensation amount is 30 minus 28 times 1 divided by 10, i.e. 0.2 ampere, i.e. the control current needs to be increased by 0.2 ampere; the calculated compensation parameter is used to correct the operating parameters of the cooling device, for example, the control current of the liquid cooling device is increased by 0.2 ampere; after correction, the parameter is continuously monitored, if the operating parameter after compensation still exceeds the preset allowable deviation threshold range, the preset allowable deviation threshold being usually ±10% of the standard value, the fault state is confirmed, a fault alarm signal is generated, the fault alarm signal including information such as the fault device, the fault parameter, the deviation value, etc., at the same time, the fault information is recorded to the log database, the log database storing contents including the fault occurrence time, the fault type, the fault parameter change curve, the treatment measures and results, facilitating subsequent fault analysis and system maintenance.

[0043] The core function of the function library optimization module is to optimize the cooling control function library based on historical cooling data, improve the matching accuracy of the execution function and the multi-dimensional cooling state vector, and further optimize the mapping relationship combined with a machine learning model. The working process of the function library optimization module is as follows: first, analyze the relevant data in the historical cooling process. The analyzed data includes the physical parameters of the material to be cooled, such as material thermal conductivity and cross-sectional structure, the multi-dimensional cooling state vector at each time, the execution function selection record, i.e., the execution function corresponding to each vector, and the cooling effect evaluation results, i.e., the temperature, deformation, and mechanical properties of the material after cooling. The analysis process uses statistical analysis methods, and the statistical period is determined according to the production batch, for example, 100 cooling cycles, to ensure that the data is representative. Then, the matching success rate of each cooling function in different vector intervals is calculated. The matching success rate is defined as the proportion of the function output that meets the target state vector requirements when using the cooling function in a certain vector interval. The target state vector requirements are that the cooling effect evaluation results meet the standards, such as the material temperature reaching the target value and the deformation being within the allowed range. The calculation method of the matching success rate is to divide the number of matching successes by the total number of uses, multiply by 100%, for example, a certain cooling function is used 100 times in a certain vector interval, 85 times meet the requirements, and the matching success rate is 85%. Next, identify the vector intervals with a matching success rate below the preset threshold. The preset threshold is determined according to the cooling quality requirements, for example, 80%. For these vector intervals, measures such as adjusting the cooling function parameters or adding new adaptive functions are taken. The adjustment range of the cooling function parameters is determined according to the success rate deviation, for example, when the matching success rate is 70%, the parameter adjustment range is ±8%. The method of adding new adaptive functions is to generate an interpolation based on the parameters of the cooling functions with good performance in adjacent intervals. The interpolation method uses linear interpolation, for example, the function parameters of the adjacent two intervals are 3 amperes and 5 amperes, respectively, and the parameter of the new function is 4 amperes. Finally, the adjusted cooling function or the new function is updated to the cooling control function library, and the vector interval corresponding to each function is updated to ensure that the function library always maintains high matching accuracy.

[0044] The function library optimization module can promote the dimension expansion of the multi-dimensional cooling state vector. The specific process is as follows: first, a profile deformation dimension is added in the multi-dimensional cooling state vector. The profile deformation dimension is an index for representing the degree of shape change of the profile in the cooling process. Its quantitative formula is the ratio of the actual deformation of the profile to the target deformation, i.e., the profile deformation dimension value is equal to the actual deformation divided by the target deformation. The actual deformation refers to the actual deformation amount of the profile in the cooling process, which is measured by a deformation monitoring device. The deformation monitoring device uses a laser displacement sensor with a measurement accuracy of ±0.01 mm and a collection frequency of 10 Hz, which can accurately measure the deformation of the profile in real time. The installation position is the key part of the profile, such as the edge or corner that is prone to deformation. The target deformation refers to the maximum allowable deformation amount set according to the subsequent processing requirements of the profile, which is calculated based on the mechanical properties of the profile material, such as elastic modulus, yield strength, and cross-sectional structure. For example, the target deformation of 6061 aluminum alloy rectangular profile is set to 0.1 mm. Then, the contribution of the profile deformation dimension is verified based on experimental test data. The experimental test data are 50 sets of cooling experimental data of different profile parameters. The contribution is evaluated using the variance contribution rate method. The variance contribution rate refers to the contribution proportion of a certain dimension to the total variance of all dimensions. The calculation method is the variance of the dimension divided by the sum of the variances of all dimensions multiplied by 100%. The variance contribution rate of the profile deformation dimension should not be less than 15%. If the requirement is met, the dimension is retained. According to the variance contribution rate of each dimension, the weight distribution of the multi-dimensional cooling state vector is adjusted. The weight distribution principle is that the higher the variance contribution rate, the greater the weight. For example, the variance contribution rate of the profile deformation dimension is 15%, and the variance contribution rates of the original temperature fluctuation coefficient, thermal distribution uniformity, and thermal conductivity coefficient are 28%, 29%, and 28%, respectively. The weights of each dimension are temperature fluctuation coefficient 0.28, thermal distribution uniformity 0.29, thermal conductivity coefficient 0.28, and profile deformation dimension 0.15. Finally, the vector database and the vector-function mapping model are updated. The vector database adds historical data of the profile deformation dimension, and the vector-function mapping model is retrained to adapt to the expanded four-dimensional vector space, ensuring that the model can accurately process the information of the new dimension.

[0045] The function library optimization module optimizes the vector-function mapping relationship using a machine learning model. The machine learning model used is a support vector machine or a decision tree model. The support vector machine is a machine learning algorithm based on statistical learning theory, which realizes data classification or regression by finding the optimal hyperplane. Here, it is used for vector-function classification matching. The key parameters of the support vector machine include the kernel function type, the penalty coefficient, the kernel function type selection radial basis function (RBF) kernel, the penalty coefficient set to 10 for balancing the fitting accuracy and generalization ability of the model, and the gamma parameter set to 0.1 for controlling the influence range of the radial basis function. The decision tree model is a machine learning algorithm based on tree structure, which realizes classification by recursively dividing the feature space. Here, it is also used for vector-function classification matching. The key parameters of the decision tree model include the maximum depth and the minimum sample split number. The maximum depth is set to 10 to avoid model overfitting, and the minimum sample split number is set to 5 to ensure that the split nodes have statistical significance. The training process of the machine learning model is as follows: first, prepare the training data set. The input of the training data set is the historical multi-dimensional cooling state vector, which needs to be processed by Z-score normalization. Z-score normalization is a method of converting data into standard normal distribution with mean 0 and standard deviation 1. The processing method is to normalize the value equal to the original value minus the mean divided by the standard deviation. The output is the corresponding execution function selection record, i.e., the label. Then, the training data set is divided into training set and test set according to the ratio of 7:3. The training set is used for model parameter learning, and the test set is used for preliminary evaluation of model performance. During model training, cross-validation method is used for performance evaluation. The 5-fold cross-validation method is selected, i.e., the training set is divided into 5 non-overlapping subsets. Four subsets are used as training data and one subset is used as validation data. After repeating 5 times, the average performance index is calculated. The performance evaluation index of the model is accuracy, which is the number of samples correctly predicted by the model divided by the total number of samples multiplied by 100%. The accuracy is required to reach a preset threshold, for example, 90%. When the model accuracy reaches the threshold, it is deployed in the system to replace the traditional matching strategy or combined with the traditional strategy to improve the matching accuracy and efficiency of the vector-function.

[0046] The core function of the cooling dynamic optimization module is to adjust the cooling parameters and the weights of the multi-dimensional cooling state vector in real time, ensure that the cooling process is always in the optimal state, and handle cooling abnormalities and record data at the same time.The working process of the cooling dynamic optimization module is as follows: firstly, continuously collecting the cooling data output by the cooling data collection module, calculating the values of cooling stability, thermal distribution uniformity, thermal conductivity coefficient and profile deformation dimension in real time according to the calculation method of the data processing and control module and the extended vector, constructing a real-time updated multi-dimensional cooling state vector, and the vector update period is consistent with the cooling data collection frequency, which is 1 second; then comparing the real-time cooling state vector with the preset ideal cooling vector, calculating the deviation index of each dimension, and the calculation method of the deviation index of each dimension is the absolute value of the difference between the actual value and the ideal value of the dimension divided by the ideal value, for example, the actual value of the temperature fluctuation coefficient is 0.0079, the ideal value is 0.008, and the deviation index is the absolute value of 0.0079 minus 0.008 divided by 0.008, which is equal to 0.0125, i.e. 1.25%; determining the cooling state abnormality or performance decline based on the deviation index of each dimension, and the determination standard is that the deviation index of any dimension is not less than 20%, if the condition is met, it is determined that the cooling state is abnormal or the performance is declined; if an abnormality is determined, automatically generating a control instruction in combination with the liquid cooling, gas cooling and air cooling device model in the cooling device calibration module, the control instruction is used to dynamically adjust the liquid flow, gas pressure, air speed and wind direction distribution, and the adjustment step is determined according to the parameter type, for example, the liquid flow adjustment step is 5 liters per minute, the gas pressure adjustment step is 0.05 megapascal, the air speed adjustment step is 1 meter per second, and the wind direction distribution adjustment step is 5 degrees, and the cooling stability, thermal distribution uniformity, thermal conductivity coefficient and profile deformation are adjusted to make the deviation index of each dimension return to the allowable range; at the same time, the weight distribution of each dimension of the multi-dimensional cooling state vector is adjusted in real time based on the historical cooling data and the experimental verification results, the historical cooling data is 1000 groups of cooling cycle data, the experimental verification results are obtained by orthogonal experiment, the orthogonal experiment design is 4 factors and 3 levels, and the factors include liquid flow, gas pressure, air speed and wind direction; the trigger condition for weight adjustment is that the cooling effect is not up to standard for 3 times in a row, and the standard for cooling effect is that the cooling effect difference index is not greater than 0.1; the weight adjustment adopts gradient descent method, which is a commonly used optimization algorithm, the gradient direction of the weight is calculated and the weight is updated in the opposite direction of the gradient, and the learning rate is set to 0.01, which is used to control the step length of weight update to avoid excessive weight fluctuation; the function matching accuracy and cooling effect are improved by adjusting the weight distribution; if the cooling state deviates from the preset threshold for a long time, the duration is set to 60 seconds, an alarm signal is generated and a preset emergency cooling function is triggered, the parameters of the emergency cooling function are consistent with those in the feedback and optimization module; at the same time, the abnormal data is recorded to the log database, including the multi-dimensional cooling state vector value, the cooling device operating parameter, the deviation index change curve and the like of the abnormal period; the abnormal data in the log database is used for subsequent analysis, and is provided to the function library optimization module as a basis for updating the cooling function parameters, so as to ensure that the function library can adapt to abnormal working conditions, and improve the overall stability and reliability of the system.

[0047] To clearly demonstrate the practical application effect of this system, the system's working process and data are explained in detail by taking the extrusion cooling process of 6061 aluminum alloy rectangular profiles as an example. This profile is used to make building door and window frames.

[0048] 1. Basic application scenario parameters: The cross-sectional dimensions of the 6061 aluminum alloy rectangular profile are 50 mm × 10 mm, and the length is 6 meters. The material parameters are thermal conductivity of 237 watts per meter Kelvin, specific heat capacity of 900 joules per kilogram Kelvin, and density of 2700 kilograms per cubic meter. The cooling target parameters are an initial temperature of 500 degrees Celsius, a target cooling temperature of 100 degrees Celsius, and a total cooling time of 300 seconds. The time distribution of the cooling rate is required to be 2 degrees Celsius per second from 500 degrees Celsius to 300 degrees Celsius in the first 100 seconds, a cooling rate of 1.5 degrees Celsius per second from 300 degrees Celsius to 150 degrees Celsius in the middle 100 seconds, and a cooling rate of 0.5 degrees Celsius per second from 150 degrees Celsius to 100 degrees Celsius in the last 100 seconds. The cooling device adopts a combination of liquid cooling, air cooling, and air cooling. The liquid cooling device is used for rapid cooling in the early stage, and the air cooling and air cooling are used for slow cooling in the middle and late stages to ensure uniform cooling.

[0049] 2. Working process of each module of the system: (1) Parameter acquisition module operation: by consulting the industry standard manual of 6061 aluminum alloy, the thermal conductivity of the material is obtained as 237 watts per meter Kelvin and the specific heat capacity is 900 joules per kilogram Kelvin; a laser profiler is used to scan the cross section of the extruded profile, and the cross-sectional structure is obtained as a rectangle with a length of 50 mm, a width of 10 mm, and a thickness of 5 mm; an infrared thermometer is used to measure the initial temperature at 6 points, including the center, edge, and four corners of the profile cross section, with an average value of 500 degrees Celsius; the target cooling temperature of 100 degrees Celsius and the cooling time requirement are obtained from the production process file; the above parameters are stored in the system database for other modules to call.

[0050] (2) cooling data acquisition module works: in the center of the section, the upper and lower edges, left and right edges of the profile 5 points installed K thermocouple, with 1 hertz frequency acquisition temperature fluctuation data; in the observation window of cooling device installed 640x512 pixel infrared thermal imager, with 0.5 hertz frequency acquisition heat distribution data; installed on the surface of the profile range 0 to 1000 watts per square meter heat flow meter, with 1 hertz frequency acquisition heat conduction data; cooling to 50 seconds, thermocouple acquisition temperature fluctuation data is 350 degrees Celsius, 348 degrees Celsius, 352 degrees Celsius, 349 degrees Celsius, 351 degrees Celsius, infrared thermal imager acquisition heat distribution data in 10 sampling point temperature value is 350 degrees Celsius, 351 degrees Celsius, 349 degrees Celsius, 352 degrees Celsius, 348 degrees Celsius, 350 degrees Celsius, 351 degrees Celsius, 349 degrees Celsius, 350 degrees Celsius, 351 degrees Celsius, heat flow meter acquisition heat flux is 1200 watts per square meter.

[0051] (3) cooling device calibration module works: the system has completed calibration before running, the correction relationship model of liquid cooling device is flow equal to 10 times the control current, the mapping relationship model of fog cooling device is airflow velocity equal to 24 times the air pressure set value, coverage equal to 0.1 times the air pressure set value, the correlation model of air cooling device is wind speed equal to 0.000002 times the fan speed square plus 0.003 times the fan speed, wind direction distribution equal to 0.002 times the fan speed plus 1, these models and corresponding calibration conditions are stored in calibration database.

[0052] (4) Data processing and control module works: the data preprocessing unit carries out mean filtering window size 5 and 3σ criterion outlier rejection on the temperature fluctuation data of 50 seconds, and the processed data is still 350 degrees Celsius, 348 degrees Celsius, 352 degrees Celsius, 349 degrees Celsius, 351 degrees Celsius; the vector calculation unit calculates the cooling stability, the average temperature is 350 degrees Celsius, the temperature standard deviation is 1.58 degrees Celsius, and the temperature fluctuation coefficient is 0.0045; the heat distribution uniformity is calculated, the sample variance is 1.44 degrees Celsius square, the overall variance is 4 degrees Celsius square, and the heat distribution uniformity is 0.64; the heat conduction coefficient is calculated, the profile surface temperature is 350 degrees Celsius, that is, 623.15 Kelvin, the cooling liquid temperature is 25 degrees Celsius, that is, 298.15 Kelvin, the temperature difference is 325 Kelvin, the heat transfer distance is 0.01 meters, and the heat conduction coefficient is 0.0369 watt per meter Kelvin, which is normalized to 0.615; a multidimensional cooling state vector is constructed, which is a vector composed of 0.0045, 0.64 and 0.0369; the function matching unit reads the vector of the last period, that is, 49 seconds, which is composed of 0.005, 0.63 and 0.037, and calculates the Euclidean distance of 0.01001, which exceeds the preset threshold value 0.01, and the matching process is started; the candidate execution function F3 is selected, and the vector interval is temperature fluctuation coefficient 0.004 to 0.005, heat distribution uniformity 0.62 to 0.66, and heat conduction coefficient 0.035 to 0.038 watt per meter Kelvin, and the preset vector is composed of 0.0045, 0.64 and 0.037; the cosine similarity is calculated as 1, F3 is selected as the target function, and the output parameters are liquid cooling control current 4 amperes, gas cooling gas pressure set value 0.6 megapascal, and air cooling fan speed 2500 revolutions per minute; the control instruction is generated in combination with the calibration model, the liquid cooling control current 4 amperes corresponds to the flow 40 liters per minute, the gas cooling gas pressure 0.6 megapascal corresponds to the gas flow speed 14.4 meters per second, and the air cooling speed 2500 revolutions per minute corresponds to the air speed 20 meters per second.

[0053] (5) Cooling device controller works: after receiving the control instruction, the analog signal is output to the device driver, the analog signal of the liquid cooling device is 16.8 milliamperes, the control current of the delivery pump is 4 amperes, and the actual flow is stabilized at 40 liters per minute through PID adjustment; the analog signal of the fog cooling device is 12 milliamperes, the gas pressure is stabilized at 0.6 megapascal; the analog signal of the air cooling device is 16 milliamperes, and the fan speed is stabilized at 2500 revolutions per minute; the running parameters are monitored in real time and fed back to the data processing and control module.

[0054] (6) Feedback and optimization module works: continuously collect cooling data and calculate multi-dimensional cooling state vector, the vector at 50 seconds is compared with the ideal cooling vector 0.004, 0.8, 0.035, the Euclidean distance is 0.16 which exceeds the allowed deviation threshold 0.1, triggering the data processing and control module to re-match; at the same time, the air pressure of the fog cooling device is 0.58 MPa which deviates from the standard range 0.3 to 0.7 MPa, the compensation parameter is calculated as 0.0083 ampere corresponding to the increase of 0.02 MPa of air pressure by calling the calibration model, and the corrected air pressure stabilizes at 0.6 MPa; the fault diagnosis unit does not monitor other abnormal parameters.

[0055] (7) Function library optimization module works: based on the historical 100 cooling cycle data, the matching success rate of F3 in the corresponding vector interval is 82% which is higher than the threshold 80%, and there is no need to adjust the parameters; at the same time, the support vector machine model is used to optimize the vector-function mapping relationship, the model input is the historical multi-dimensional cooling state vector, and the output is the execution function selection record, the training set and test set ratio is 7 to 3, and the accuracy rate after 5-fold cross-validation is 92% which reaches the threshold 90%, and it has been deployed to the system.

[0056] (8) Cooling dynamic optimization module works: continuously collect cooling data, the real-time vector at 50 seconds is compared with the ideal vector, the thermal distribution uniformity deviation index is 20% which is judged as abnormal, and the control command is generated to adjust the liquid cooling nozzle angle to make the cooling liquid spray more uniform; based on 1000 groups of historical data and orthogonal test results, the weight of each dimension is adjusted, the temperature fluctuation coefficient weight is 0.25, the thermal distribution uniformity weight is 0.3, the thermal conductivity coefficient weight is 0.25, and the profile deformation amount dimension weight is 0.2; no cooling state deviation threshold is monitored, and no emergency function is triggered.

[0057] 3. Application effect: through the control of the system, the 6061 aluminum alloy rectangular profile is successfully cooled from 500 degrees Celsius to 100 degrees Celsius within 300 seconds, the temperature fluctuation coefficient during the cooling process is always not greater than 0.01, i.e. 1%, the thermal distribution uniformity is always not less than 0.8, i.e. 80%, and the deformation amount of the profile after cooling is 0.08 mm which is less than the target deformation amount 0.1 mm, meeting the subsequent stretching processing requirements.

[0058] The above is only a preferred embodiment of the present application, and the protection scope of the present application is not limited to the above-mentioned embodiments, any technical solution falling within the concept of the present application shall fall within the protection scope of the present application. It should be noted that for ordinary technical personnel in the technical field, some improvements and decorations without departing from the principles of the present application shall also be considered as the protection scope of the present application.

Claims

1. An extrusion cooling online detection and feedback correction system, characterized by: include A parameter acquisition module is used to obtain the physical parameters of the metal profile to be cooled, including the material thermal conductivity, specific heat capacity, cross-sectional structure, initial temperature, target cooling temperature, and cooling time requirements, including the total cooling time and the time distribution requirements of the cooling rate; A cooling data acquisition module is used to collect cooling data during the cooling process, wherein the cooling data includes temperature fluctuation data, heat distribution data, and heat conduction data; The cooling device calibration module is used to test the output characteristics of liquid cooling devices, mist cooling devices, and air cooling devices, and to establish a corresponding relationship model between the control instructions and the actual output parameters of the cooling device; A data processing and control module calculates cooling stability, heat distribution uniformity, and heat transfer coefficient based on cooling data, constructs a multi-dimensional cooling state vector, matches the multi-dimensional cooling state vector updated in real time with an execution function in a preset cooling control function library, and generates control instructions; The cooling device controller is used to receive the control instruction to adjust the operating parameters of the cooling device.

2. The extrusion cooling online detection and feedback correction system according to claim 1, characterized in that: The cooling device calibration module includes: Liquid cooling calibration unit, used to establish a correction relationship model between the liquid cooling device control instructions and the coolant flow and temperature output through least squares fitting; An air cooling calibration unit is used to establish a mapping relationship model between the air pressure setting value of the mist cooling device and the air flow velocity and coverage range through regression analysis; Air cooling calibration unit, used to construct a correlation model between the fan speed of the air cooling device and the wind speed and wind direction distribution through curve fitting; The calibration database is used to store the correction relationship model, the mapping relationship model and the association model.

3. The extrusion cooling online detection and feedback correction system according to claim 1 is characterized in that: The data processing and control module includes: A data preprocessing unit, used to filter the cooling data and remove outliers; The vector calculation unit is used to calculate the index value of each dimension of the vector based on a preset quantization formula; wherein, The cooling stability is quantified using the temperature fluctuation coefficient, which is calculated as the ratio of the temperature standard deviation to the average temperature; The heat distribution uniformity is quantified using the temperature distribution uniformity metric, which is calculated as 1 minus the ratio of the temperature sample variance to the overall variance.

4. The extrusion cooling online detection and feedback correction system according to claim 3 is characterized in that: The data processing and control module further includes a function matching unit, which is configured to: Read the latest multi-dimensional cooling state vector and calculate its Euclidean distance with the vector of the previous cycle; If the Euclidean distance exceeds a preset threshold, or the latest vector exceeds the vector interval corresponding to the currently executed function, the matching process is started; A matching strategy based on a preset vector interval is used to screen candidate execution functions from the cooling control function library, and the cosine similarity between the multi-dimensional cooling state vector and the preset vector of each candidate function is calculated; The execution function with the highest cosine similarity is selected as the objective function, and the output parameters of the objective function are converted into control instructions in combination with the model in the cooling device calibration module.

5. The extrusion cooling online detection and feedback correction system according to claim 1, characterized in that: The module further includes a feedback and optimization module, wherein the feedback and optimization module is configured to: Continuously collecting cooling data and calculating a multi-dimensional cooling state vector, and comparing it with an ideal cooling vector, wherein the ideal cooling vector is a target state vector calculated based on cooling target parameters; The Euclidean distance between the two is calculated as the cooling effect difference indicator; If the difference indicator is within the preset allowable deviation threshold, the parameters of the currently executed function are fine-tuned and optimized; If the difference index exceeds the preset allowable deviation threshold range, the data processing and control module is triggered to re-match the function; If it is monitored that the operating parameters of the cooling device continue to deviate from the standard range, an alarm signal will be generated and the preset emergency cooling function will be called.

6. The extrusion cooling online detection and feedback correction system according to claim 5, characterized in that: The feedback and optimization module further includes a fault diagnosis unit, which is configured to: Monitor cooling unit operating parameters and compare them with pre-set standard ranges; If the parameters deviate, the corresponding model in the calibration database is called to calculate the compensation parameters and used to correct the operating parameters; If the operating parameters after compensation still exceed the preset allowable deviation threshold range, it is confirmed as a fault state, a fault alarm signal is generated and recorded in the log database.

7. The extrusion cooling online detection and feedback correction system according to claim 2, characterized in that: It also includes a function library optimization module, which is configured to: Analyze profile parameters, multi-dimensional cooling state vectors, execution function selection records and cooling effect evaluation results during historical cooling processes; Counting the matching success rate of each cooling function in different vector intervals, where the matching success rate is defined as the ratio of the function output that meets the target state vector requirement; Identify the vector interval where the matching success rate is lower than the preset threshold, and adjust the parameters of the cooling function or add a new adaptation function within this interval.

8. The extrusion cooling online detection and feedback correction system according to claim 7, characterized in that: A new dimension of profile deformation is added to the multidimensional cooling state vector. Its quantitative formula is the ratio of the actual deformation of the profile to the target deformation. The actual deformation is measured by the deformation monitoring device. The contribution of the profile deformation dimension is verified based on experimental test data, and the weight distribution of the multi-dimensional cooling state vector is adjusted; The vector database and vector-function mapping model are updated to adapt to the expanded multi-dimensional cooling state vector space.

9. The extrusion cooling online detection and feedback correction system according to claim 8, characterized in that: The function library optimization module uses a machine learning model to optimize the vector-function mapping relationship, and the machine learning model is a support vector machine or a decision tree model; The machine learning model is trained with a historical multi-dimensional cooling state vector as input and an execution function selection record as output; The model is evaluated for performance through cross-validation and deployed into the system after the accuracy reaches a preset threshold.

10. The extrusion cooling online detection and feedback correction system according to claim 1, characterized in that: The cooling dynamic optimization module is further included, and the cooling dynamic optimization module is configured to: Continuously collecting the cooling data, and calculating the cooling stability, heat distribution uniformity, heat transfer coefficient and profile deformation dimension, to construct a multi-dimensional cooling state vector that is updated in real time; Compare the real-time cooling state vector with the preset ideal cooling vector, calculate the deviation index of each dimension, and determine whether the cooling state is abnormal or the performance is degraded based on the deviation index; If an abnormality is detected, the system automatically generates control instructions based on the liquid cooling, air cooling, and wind cooling device models in the cooling device calibration module to dynamically adjust the liquid flow rate, air pressure, wind speed, and wind direction distribution to optimize cooling stability, heat distribution uniformity, heat transfer coefficient, and profile deformation. Based on historical cooling data and experimental verification results, the weight distribution of each dimension of the multi-dimensional cooling state vector is adjusted in real time to improve function matching accuracy and cooling effect; When the cooling state continues to deviate from the preset threshold, an alarm signal is generated and the preset emergency cooling function is triggered, and the abnormal data is recorded in the log database for subsequent analysis and function library optimization module to update the cooling function parameters.

Images