Section steel self-adaptive straightening system based on multi-physics field coupling and prediction compensation method

Abstract

The invention relates to a profile steel self-adaptive straightening system based on multi-physics field coupling and a prediction compensation method, and belongs to the technical field of metal material machining. Aiming at the problems of contradiction between precision and efficiency, large residual stress and poor universality in traditional section steel straightening, a three-stage temperature control process is combined with a pressure distribution model of an elastic modulus temperature function and a cross section quadratic moment, real-time defect detection is realized through high-density laser array scanning and a ResNet / MobileNet classifier, and a 500ms ultra-short period closed-loop control system is constructed. A hybrid expert model MoE is adopted to predict compensation parameter offset, so that the residual stress after straightening is 1t; the precision of the head of the section steel is less than or equal to 50MPa, the precision of the head of the section steel is less than or equal to 0.5 mm / 1000mm, the specification of 100mm * 50mm-1750mm * 550mm is covered, the energy consumption is reduced by more than 20%, and the efficient and precise straightening

Description

Technical Field

[0001] The invention belongs to the technical field of metal material processing, and relates to a steel section adaptive straightening system based on multi-physical field coupling and a prediction compensation method. Background Art

[0002] Steel sections are the core of major projects such as green prefabricated buildings, large steel structure facilities, high-rise buildings, bridges, etc. The main products include large, medium and small steel sections. According to data from 2023, the output of steel sections is 73.8 million tons, accounting for 7.25% of my country's crude steel output. With the continuous expansion of the market demand for steel sections and the gradual expansion of the size range of single-line steel sections, steel production companies have higher and higher requirements for production efficiency, and downstream users have higher and higher requirements for product flatness. Straightening, as an important finishing process in the production process of steel sections, directly determines the final flatness of the product.

[0003] Existing steel straightening technology is mainly based on the principle of mechanical deformation, which eliminates internal stress or deformation of the material by applying external force. According to the technical implementation method, the existing straightening systems and methods can be divided into the following categories:

[0004] 1) Roller straightening technology

[0005] Using multiple sets of staggered straightening rollers to continuously bend and straighten the steel has the advantages of high efficiency and mechanization. For example, the nine-roller straightening machine can adapt to different specifications of steel by adjusting the rollers, but it relies on manual experience to adjust the rollers and lacks flexibility.

[0006] 2) Pressure straightening technology

[0007] The reverse bending force is applied by hydraulic cylinder or mechanical pressure, which is suitable for local correction of large steel sections, but it requires multiple operations and has low efficiency. Pressure straightening machines are often used as supplementary correction equipment and require manual operation, which is difficult to meet the needs of continuous production.

[0008] 3) Combined straightening technology

[0009] Combining multi-roller pre-straightening and pressure final straightening to improve straightness (such as H-beam straightening device), but the equipment structure is complex and the energy consumption is high. Some patents propose linkage control of pinch rollers and straightening machines, and adjusting the conveying speed through variable frequency motors, but too many monitoring units lead to increased maintenance costs.

[0010] In response to various problems occurring during the straightening process of profiled steel, experts and scholars have proposed different straightening systems and methods. The patent "Straightening Process and Structure for Profiled Steel (CN 114309074B)" utilizes the roll matching technology of a nine-roll straightening machine to improve the straightening efficiency. However, the roll matching process relies on manual experience and is difficult to quickly adapt to new cross-section profiled steel. The patent "Straightening Device for Aluminum Profile Processing (CN 222491593U)" proposes a preheating component to soften the material to reduce straightening damage, but it does not solve the universality problem of multiple specifications of profiled steel during the straightening process. The patent "Multi-functional Machine for Bending and Straightening Reinforcing Bars (CN222491890U)" adjusts the distance between straightening rolls through a sliding sleeve to improve the applicability to reinforcing bars, but it does not involve the targeted design of complex cross-sections of profiled steel. The patent "Automatic Profiled Steel Straightening Equipment (CN119281868A)" uses a displacement sensor and an equipment controller to achieve automatic straightening, but it relies on high-precision sensors, has a high cost, and insufficient anti-interference ability.

[0011] During the production process of profiled steel straightening, the following problems still exist: There is a contradiction between straightening accuracy and efficiency. Although a multi-roll straightening machine has high efficiency, the adjustment of the distance between straightening rolls relies on manual experience and it is difficult to achieve high-precision adaptation. Pressure straightening requires repeated operations, has a low degree of automation, and cannot meet the requirements of continuous production. The problem of residual stress. During the straightening process, the calculation of residual stress is inaccurate, resulting in springback after straightening. The problems of material damage and energy consumption. During the traditional straightening process, the material is prone to surface scratches or internal cracks due to plastic deformation. Although the preheating technology can alleviate the damage, the energy consumption increases significantly. Insufficient universality. The straightening device designed for a specific cross-section profiled steel (such as H-shaped steel) is difficult to adapt to other complex cross-section profiles, and the equipment reuse rate is low. Most of the existing patent technologies focus on the optimization of local structures and lack a systematic control method for straightening process parameters. In view of the problems existing in the straightening process of the current profiled steel production line, the present invention provides a profiled steel adaptive straightening system and a prediction compensation method based on multi-physical field coupling, which can effectively improve the flatness and surface quality of profiled steel after straightening, while improving the stability of straightening production, improving the straightening adaptability and operation rate, increasing the average hourly output, improving the automation and intelligent level of the production line, reducing the operation cost, and improving the economic benefits. Summary of the Invention

[0012] In view of this, the purpose of the present invention is to provide a profiled steel adaptive straightening system and a prediction compensation method based on multi-physical field coupling. This method has a low construction cost, is easy to realize industrial production, has a low operation cost, and the produced profiled steel products have high flatness and excellent surface quality, with obvious economic benefits.

[0013] To achieve the above object, the present invention provides the following technical solutions:

[0014] A profiled steel adaptive straightening system based on multi-physical field coupling, comprising:

[0015] A multi-stage temperature control module is used to perform three-stage temperature control on the section steel, namely high-temperature straightening, medium-temperature straightening, and low-temperature straightening. The high-temperature straightening temperature is 600°C to 900°C, the medium-temperature straightening temperature is 400°C to 600°C, and the low-temperature straightening temperature is 0°C to 100°C. Moreover, the module integrates a temperature sensor and a PID algorithm to dynamically compensate for the surface temperature of the straightening rolls, with a control error less than ±5°C;

[0016] An adaptive pressure distribution module is configured to calculate the pressure distribution values at each straightening point based on the function of the elastic modulus of the section steel material varying with temperature, the second moment of area at each straightening point, the dynamic correction coefficient of material viscosity, and the deformation rate influence factor function, and update the pressure parameters every 500 ms;

[0017] A high-precision defect detection module includes a high-density laser array and a deep learning classifier. The laser array generates 3D topography point cloud data of the section steel at a sampling rate of ≥2000 points / second, and the classifier is based on the ResNet or MobileNet architecture and uses Focal Loss to optimize defect classification;

[0018] A dynamic feedback control module is used to fuse the data of pressure, temperature, and deformation sensors, construct a closed-loop feedback system with a 500 ms control cycle, and predict the deviation trend of straightening parameters in combination with a Mixture of Experts (MoE) model to achieve dynamic compensation.

[0019] Furthermore, in the high-temperature straightening stage, an induction heat compensation device is used to adjust the temperature of the web and flange of the section steel to the set range; in the low-temperature straightening stage, a semiconductor refrigeration technology is used to control the temperature of the straightening rolls, and the cooling stepping speed is dynamically adjusted based on a temperature model.

[0020] Furthermore, the pressure distribution model of the adaptive pressure distribution module satisfies:

[0021]

[0022] P i is the pressure value at the i-th straightening point, E(T) is the elastic modulus temperature function, I i is the second moment of area, μ is the viscosity correction coefficient, ∑W j is the normalized weight coefficient, is a correction factor function including Poisson's ratio and deformation rate.

[0023] Furthermore, the closed-loop feedback system converges the residual stress after straightening to <50 MPa, and controls the head accuracy of the section steel ≤0.5 mm / 1000 mm and the middle accuracy ≤0.3 mm / 1000 mm.

[0024] Furthermore, the Mixture of Experts (MoE) model is trained with historical data, capable of predicting the offset of straightening parameters 30 seconds in advance, and adjusting the roll gap and roll speed through a dynamic compensation strategy.

[0025] Furthermore, the system is adapted to cantilever or gantry multi-roll straightening machines, covering H-beams, rails, and high-strength steels with specifications ranging from 100mm×50mm to 1750mm×550mm, and the exit speed reaches 10.0m / s.

[0026] A method for predicting and compensating the straightening of profiled steel based on multi-physical field coupling includes the following steps:

[0027] Step 1: The profiled steel is successively subjected to high-temperature straightening at 600°C - 900°C, medium-temperature straightening at 400°C - 600°C, and low-temperature straightening at 0°C - 100°C, and the temperature of the straightening rolls is compensated in real time through the PID algorithm at each stage, with the error controlled within ±5°C.

[0028] Step 2: According to the elastic modulus temperature function, the second moment of the cross-section, the viscous correction coefficient, and the deformation rate of the profiled steel, calculate the pressure distribution values at each straightening point, and update the pressure parameters through real-time optimization with a 500ms cycle.

[0029] Step 3: Use a high-density laser array to scan the surface of the profiled steel to generate three-dimensional topography point cloud data, and identify defects through a deep learning classifier. The classifier uses Focal Loss to optimize multi-class classification.

[0030] Step 4: Integrate the feedback data from the pressure, temperature, and deformation sensors to construct a 500ms closed-loop control cycle. Combine the Mixture of Experts (MoE) model to predict the offset trend of straightening parameters, and dynamically adjust the roll gap and roll speed to make the residual stress < 50MPa.

[0031] Furthermore, in Step 1, if the straightness fails to meet the standard after straightening in a certain stage, recalculate the pressure distribution parameters for this stage and subsequent stages, and apply the corrected parameters to the next profiled steel.

[0032] Furthermore, in Step 3, the deep learning classifier performs image enhancement and sample expansion on the defect data, and is trained based on the ResNet or MobileNet architecture, with the classification response time ≤ 100ms.

[0033] Furthermore, in Step 4, the Mixture of Experts (MoE) model generates compensation parameters through the real-time collected deformation rate, temperature gradient, and historical process data, and the prediction time window is 30 seconds.

[0034] The beneficial effects of the present invention are as follows:

[0035] 1) Multi-stage temperature control and adaptive pressure distribution. The system realizes precise control of the elastic-plastic deformation of materials in different temperature ranges through a three-stage straightening process of high temperature (600°C - 900°C), medium temperature (400°C - 600°C), and low temperature (0°C - 100°C), combined with an adaptive pressure distribution algorithm based on the section moment of inertia, significantly reducing residual stress and improving straightening uniformity.

[0036] 2) Real-time closed-loop feedback and high-precision control. A closed-loop feedback system with a 500ms ultra-short control cycle is adopted, integrating a high-density laser array (≥2000 points / second sampling rate) and a deep learning defect classifier (based on ResNet and MobileNet architectures) to monitor the deformation of the section steel in real time and dynamically compensate, ensuring that the residual stress after straightening is <50MPa, the head accuracy is ≤0.8mm / 1000mm, and the middle part accuracy of the H-section steel is ≤0.6mm / 1000mm.

[0037] 3) Intelligent defect detection and modular design. The deep learning classifier realizes rapid defect classification through data augmentation and Focal Loss optimization. Combined with the modular roll system design, it can be adapted to cantilever / gantry multi-roll straightening machines and support the efficient processing of complex-section profiles such as H-section steel and rails.

[0038] 4) Wide applicability and high production efficiency. It covers wide / medium / narrow flange H-section steel, carbon structural steel, and high-strength steel with specifications ranging from 100mm×50mm to 1750mm×550mm, and the export speed reaches 10.0m / s, meeting the requirements of large-scale continuous production.

[0039] 5) Deep integration of theoretical model and engineering practice. The pressure distribution model introduces the elastic modulus temperature function E(T), the second moment of the section Ii, and the viscous correction factor. Combined with the experimentally calibrated parameters, it effectively solves the conservative error problem of the traditional model and reduces energy consumption by more than 20%.

[0040] 6) Comprehensive multi-physical field coupling control, intelligent defect detection, and rapid response mechanism are used to achieve efficient and precise straightening of complex-section steel, improving the product quality consistency and the automation and intelligence level of the production line.

[0041] Other advantages, objectives, and features of the present invention will be described to some extent in the subsequent specification, and to some extent, will be obvious to those skilled in the art based on the study of the following text, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following specification. Description of the Drawings

[0042] In order to make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be described in detail preferably with reference to the accompanying drawings, where:

[0043] Figure 1 is the process flow chart of the present invention;

[0044] Figure 2 is the control system architecture diagram of the present invention;

[0045] Figure 3 is the flow chart of the deep learning defect classifier of the present invention. Specific embodiments

[0046] The following uses specific specific examples to illustrate the implementation manners of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific implementation manners. Various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the drawings provided in the following embodiments only schematically illustrate the basic concept of the present invention. Without conflict, the following embodiments and the features in the embodiments can be combined with each other.

[0047] Among them, the drawings are only for illustrative purposes, showing only schematic diagrams, not physical diagrams, and cannot be understood as a limitation to the present invention; in order to better illustrate the embodiments of the present invention, some components in the drawings will be omitted, enlarged or reduced, which does not represent the size of the actual product; for those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0048] In the drawings of the embodiments of the present invention, the same or similar reference numerals correspond to the same or similar components; in the description of the present invention, it should be understood that if there are terms such as "upper", "lower", "left", "right", "front", "rear", etc. indicating the orientation or positional relationship, they are based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, the terms describing the positional relationship in the drawings are only for illustrative purposes and cannot be understood as a limitation to the present invention. For those of ordinary skill in the art, the specific meanings of the above terms can be understood according to specific circumstances.

[0049] A section steel adaptive straightening system and prediction compensation method based on multi-physical field coupling. The system and the straightening method include:

[0050] (1) Multi-stage temperature control module

[0051] (1.1) Three-stage temperature control process: High-temperature straightening (600°C - 900°C): Utilize the high-temperature plastic deformation characteristics of materials to reduce the straightening resistance; Medium-temperature straightening (400°C - 600°C): Balance elastic-plastic deformation and residual stress; Low-temperature straightening (0°C - 100°C): Combine semiconductor high-efficiency refrigeration technology to achieve precise temperature control and suppress cold deformation springback.

[0052] (1.2) Dynamic temperature compensation: Integrate temperature sensors and PID algorithms to adjust the surface temperature of the straightening rolls in real time, with an error < ±5°C.

[0053] (2) Adaptive pressure distribution module

[0054] (2.1) Core algorithm pressure distribution model: The pressure distribution model during straightening is:

[0055]

[0056] Among them, E(T), the elastic modulus of the section steel material; I i , the second moment of the cross-section at the i-th straightening point; μ: Material viscosity dynamic correction coefficient; ∑W j , weight coefficient, used for normalizing pressure distribution, reflecting the relative importance or constraint conditions of different units; The influence factor function of Poisson's ratio and deformation rate on the straightening of section steel during straightening.

[0057] (2.2) Real-time pressure optimization: Update the pressure distribution parameters every 500 ms to achieve high-precision straightening.

[0058] (3) High-precision defect detection module

[0059] (3.1) High-density laser array: Adopt laser scanning technology with a sampling rate of 2000 points / second to generate 3D topography point cloud data of section steel;

[0060] (3.2) Deep learning classifier: Data preprocessing, image enhancement and sample expansion to improve the generalization ability of the model; Model architecture, build a classification network based on ResNet / MobileNet, and optimize classification with Focal Loss; Real-time optimization: High response for defect recognition.

[0061] (4) Dynamic feedback control module

[0062] (4.1) Closed-loop control mechanism: Through the fusion of multiple sensors of pressure, temperature, and deformation, construct a 500 ms ultra-short cycle feedback closed-loop, with the residual stress convergence value < 50 MPa;

[0063] (4.2) Prediction Compensation Strategy: Combining historical data with the MoE (Mixture of experts) model, predict the deviation trend of the straightening parameters 30 seconds in advance, and dynamically compensate for the accuracy deviation.

[0064] One of the special cases of the adaptive straightening system and prediction compensation method for profiled steel based on multi-physical field coupling is as follows:

[0065] I. Multi-stage Temperature Control Module

[0066] (1) Three-stage temperature control process: High-temperature straightening (600°C - 900°C): Utilize the high-temperature plastic deformation characteristics of the material to reduce the straightening resistance; Medium-temperature straightening (400°C - 600°C): Balance elastic-plastic deformation and residual stress; Low-temperature straightening (0°C - 100°C): Combine semiconductor high-efficiency refrigeration technology to achieve precise temperature control and suppress cold deformation springback.

[0067] (2) Dynamic temperature compensation: Integrate temperature sensors and PID algorithms to adjust the surface temperature of the straightening rolls in real time, with an error of <±5°C.

[0068] II. Adaptive Pressure Distribution Module

[0069] (1) Core algorithm pressure distribution model: The pressure distribution model during straightening is:

[0070]

[0071] Among them, E(T) is the elastic modulus of the profiled steel material; I i is the second moment of the cross-section at the i-th straightening point; μ is the dynamic correction coefficient of material viscosity; ∑W j is the weight coefficient, used for normalizing the pressure distribution, reflecting the relative importance or constraint conditions of different units; is the influence factor function of Poisson's ratio and deformation rate on the straightening of profiled steel during straightening.

[0072] (2) Real-time pressure optimization: Update the pressure distribution parameters every 500 ms to achieve high-precision straightening.

[0073] III. High-precision Defect Detection Module

[0074] (1) High-density laser array: Adopt laser scanning technology with a sampling rate of 2000 points / second to generate 3D shape point cloud data of profiled steel;

[0075] (2) Deep learning classifier: Data preprocessing, image enhancement and sample expansion to improve the generalization ability of the model; Model architecture, build a classification network based on ResNet / MobileNet, and optimize the classification with Focal Loss; Real-time optimization: High response for defect recognition.

[0076] IV. Dynamic Feedback Control Module

[0077] (1) Closed-loop control mechanism: By fusing multiple sensors for pressure, temperature, and deformation, a 500ms ultra-short cycle feedback closed-loop is constructed, with the residual stress convergence value < 50MPa;

[0078] (2) Prediction compensation strategy: Combining historical data with the MoE (Mixture of experts) mixed expert model, the deviation trend of straightening parameters is predicted 30 seconds in advance, and the accuracy deviation is dynamically compensated.

[0079] Through multi-stage temperature regulation and adaptive pressure distribution, precise control of the elastic-plastic deformation of materials in different temperature ranges is achieved, significantly reducing residual stress and improving straightening uniformity. Through real-time closed-loop feedback and high-precision control, the deformation of the section steel is monitored in real time and dynamically compensated to ensure high residual stress, head accuracy, and middle part accuracy of the section steel after straightening. Through intelligent defect detection and modular design, defects can be quickly classified, and it can be adapted to cantilever / gantry multi-roll straightening machines, supporting the efficient processing of complex cross-section profiles such as H-beams and rails. It can cover wide / medium / narrow flange H-beams, carbon structural steel, and high-strength steel with specifications ranging from 100mm×50mm to 1750mm×550mm, meeting the needs of large-scale continuous production. The theoretical model is deeply integrated with engineering practice, effectively solving the problem of conservative error in traditional models and reducing energy consumption by more than 20%. Combining multi-physical field coupling control, intelligent defect detection, and rapid response mechanism, efficient and precise straightening of complex section steel is achieved, improving the consistency of product quality and the automation and intelligence level of the production line.

[0080] Example 1: Straightening example of small H-beam (taking Q355B, H150×150×5×7 as an example)

[0081] Refer to Figures 1 to 3 Table 1 and Table 2.

[0082] Table 1 Experimental data table of the present invention (taking Q355B as an example)

[0083]

[0084] Table 2 Comparison table of straightness of products between traditional straightening and the present invention (taking Q355B, H350×250×9×14 as an example)

[0085]

[0086]

[0087] According to the present invention, the straightening process sequence includes high-temperature straightening, medium-temperature straightening, and low-temperature straightening. The specific process is as follows:

[0088] (1) High-temperature straightening (600°C - 900°C):

[0089] (1) Detect the entrance shape before high-temperature straightening, and use a high-precision defect detection module to obtain the cross-sectional dimensions and flatness data before high-temperature straightening;

[0090] (2) Based on the cross-sectional dimensions and flatness data before straightening, use the core algorithm pressure distribution model to calculate that the reduction for this pass of straightening is 20.2 mm;

[0091] (3) During the straightening process, continuously detect the temperatures of the web and flange of the section steel to ensure that the straightening temperature is 850 ± 20°C. If the temperature does not meet the standard, start the induction preheating device before straightening for preheating;

[0092] (4) After straightening, continuously detect the cross-sectional dimensions and flatness data of the section steel. If the head accuracy of the H-section steel: ≤ 2.5 mm / 1000 mm, and the middle accuracy of the H-section steel: ≤ 2 mm / 1000 mm, then enter the medium-temperature straightening process; if the flatness does not meet the standard after hot straightening, repeat the process of step (2) in the above process (1), recalculate the straightening specification, and use this specification to straighten the next piece of steel.

[0093] (2) Medium-temperature straightening (400°C - 600°C):

[0094] (1) Based on the cross-sectional dimensions and flatness data after high-temperature straightening, use the core algorithm pressure distribution model to calculate that the reduction for this pass of straightening is 14.5 mm;

[0095] (2) During the straightening process, continuously detect the temperatures of the web and flange of the section steel to ensure that the straightening temperature is 650 ± 15°C. If the temperature does not meet the standard, start the induction preheating device before straightening for preheating;

[0096] (4) After straightening, continuously detect the cross-sectional dimensions and flatness data of the section steel. If the head accuracy of the H-section steel: ≤ 1.5 mm / 1000 mm, and the middle accuracy of the H-section steel: ≤ 1.0 mm / 1000 mm, then enter the cooling bed for cooling, and then enter the low-temperature straightening process; if the flatness does not meet the standard after hot straightening, repeat the process of step (1) in the above process (2), recalculate the straightening specification, and use this specification to straighten the next piece of steel.

[0097] (3) Low-temperature straightening (0°C - 100°C):

[0098] (1) Detect the entrance shape before low-temperature straightening, and use a high-precision defect detection module to obtain the cross-sectional dimensions and flatness data before low-temperature straightening;

[0099] (2) Based on the cross-sectional dimensions and flatness data before straightening, use the core algorithm pressure distribution model to calculate that the reduction for this multi-pass straightening is 12 / 10 / 8 / 5 / 3 / 2 / 1 / 1 mm;

[0100] (3) During the straightening process, the temperatures of the web and flange of the section steel are detected in real time to ensure that the straightening temperature is 45 ± 3 °C. If the temperature does not meet the standard, the cooling step speed is adjusted according to the temperature calculation model;

[0101] (4) After straightening, the section size and flatness data of the section steel are detected in real time. If the head accuracy of the H-section steel after sizing and multiple-length sizing is ≤ 0.5 mm / 1000 mm, and the middle part accuracy of the H-section steel is ≤ 0.3 mm / 1000 mm, and the calculated value of the residual stress < 50 MPa, then it enters the medium-temperature straightening process; if the flatness does not meet the standard after hot straightening, the process of the pressure distribution model in the above processes (1), (2) and (3) is recalculated, the straightening schedule is recalculated, and this schedule is used for straightening in the next piece of steel.

[0102] Example 2: Straightening example of medium-sized H-section steel (taking Q355B, H350×250×9×14 as an example)

[0103] Refer to Figures 1 to 3 Tables 1 and 2, according to the present invention, the straightening process sequence includes: high-temperature straightening, medium-temperature straightening and low-temperature straightening. The specific process is as follows:

[0104] 1) High-temperature straightening (600 °C - 900 °C):

[0105] (1) Before high-temperature straightening, the inlet shape is detected, and the section size and flatness data before high-temperature straightening are obtained by using a high-precision defect detection module;

[0106] (2) Based on the section size and flatness data before straightening, the reduction of this pass of straightening is calculated to be 15.6 mm by using the core algorithm pressure distribution model;

[0107] (3) During the straightening process, the temperatures of the web and flange of the section steel are detected in real time to ensure that the straightening temperature is 900 ± 20 °C. If the temperature does not meet the standard, the pre-straightening induction heating device is started for heat compensation;

[0108] (4) After straightening, the section size and flatness data of the section steel are detected in real time. If the head accuracy of the H-section steel is ≤ 2.5 mm / 1000 mm, and the middle part accuracy of the H-section steel is ≤ 2 mm / 1000 mm, then it enters the medium-temperature straightening process; if the flatness does not meet the standard after hot straightening, the process of step (2) in the above process (1) is repeated, the straightening schedule is recalculated, and this schedule is used for straightening in the next piece of steel

[0109] 2) Medium-temperature straightening (400 °C - 600 °C):

[0110] (1) Based on the section size and flatness data after high-temperature straightening, the reduction of this pass of straightening is calculated to be 11.9 mm by using the core algorithm pressure distribution model;

[0111] (2) During the straightening process, the temperatures of the web and flange of the section steel are detected in real time to ensure that the straightening temperature is 750 ± 15 °C. If the temperature does not meet the standard, the induction preheating device before straightening is started for preheating;

[0112] (4) After straightening, the section size and flatness data of the section steel are detected in real time. If the accuracy of the head of the H-section steel: ≤ 1.5 mm / 1000 mm, and the accuracy of the middle part of the H-section steel: ≤ 1.0 mm / 1000 mm, it enters the cooling bed for cooling, and then enters the low-temperature straightening process; if the flatness does not meet the standard after hot straightening, repeat the process (1) in the above process 2), recalculate the straightening procedure, and use this procedure to straighten the next piece of steel

[0113] 3) Low-temperature straightening (0 °C - 100 °C):

[0114] (1) Before low-temperature straightening, the inlet shape is detected, and the high-precision defect detection module is used to obtain the section size and flatness data before low-temperature straightening;

[0115] (2) Based on the section size and flatness data before straightening, the core algorithm pressure distribution model is used to calculate that the reduction for multi-pass straightening is 10 / 8 / 7 / 4 / 3 / 2 / 1 / 1 mm;

[0116] (3) During the straightening process, the temperatures of the web and flange of the section steel are detected in real time to ensure that the straightening temperature is 55 ± 3 °C. If the temperature does not meet the standard, the cooling step speed is adjusted according to the temperature calculation model;

[0117] (4) After straightening, the section size and flatness data of the section steel are detected in real time. If after sizing and multiple-length cutting, the accuracy of the head of the H-section steel: ≤ 0.5 mm / 1000 mm, the accuracy of the middle part of the H-section steel: ≤ 0.3 mm / 1000 mm, and the calculated value of residual stress < 50 MPa, it enters the medium-temperature straightening process; if the flatness does not meet the standard after hot straightening, recalculate the process of the pressure distribution model in the above processes 1), 2) and 3), recalculate the straightening procedure, and use this procedure to straighten the next piece of steel.

[0118] Example 3: Straightening example of large H-section steel (taking Q355B, H600×300×12×17 as an example)

[0119] Refer to Figures 1 to 3 Tables 1 and 2, according to the present invention, the straightening process sequence includes: high-temperature straightening, medium-temperature straightening and low-temperature straightening. The specific process is as follows:

[0120] 1) High-temperature straightening (600 °C - 900 °C):

[0121] (1) Before high-temperature straightening, the inlet shape is detected, and the high-precision defect detection module is used to obtain the section size and flatness data before high-temperature straightening;

[0122] (2) Based on the cross-sectional dimensions and flatness data before straightening, the reduction for this pass of straightening is calculated to be 13.1 mm using the core algorithm pressure distribution model;

[0123] (3) During the straightening process, the temperatures of the web and flange of the section steel are detected in real time to ensure that the straightening temperature is 920 ± 20 °C. If the temperature does not meet the standard, the induction preheating device before straightening is started for preheating;

[0124] (4) After straightening, the cross-sectional dimensions and flatness data of the section steel are detected in real time. If the head accuracy of the H-section steel: ≤ 2.5 mm / 1000 mm, and the middle part accuracy of the H-section steel: ≤ 2 mm / 1000 mm, it enters the medium-temperature straightening process; if the flatness does not meet the standard after hot straightening, repeat the process of step (2) in the above process (1), recalculate the straightening specification, and use this specification to straighten the next piece of steel

[0125] 2) Medium-temperature straightening (400 °C - 600 °C):

[0126] (1) Based on the cross-sectional dimensions and flatness data after high-temperature straightening, the reduction for this pass of straightening is calculated to be 9.5 mm using the core algorithm pressure distribution model;

[0127] (2) During the straightening process, the temperatures of the web and flange of the section steel are detected in real time to ensure that the straightening temperature is 750 ± 15 °C. If the temperature does not meet the standard, the induction preheating device before straightening is started for preheating;

[0128] (4) After straightening, the cross-sectional dimensions and flatness data of the section steel are detected in real time. If the head accuracy of the H-section steel: ≤ 1.5 mm / 1000 mm, and the middle part accuracy of the H-section steel: ≤ 1.0 mm / 1000 mm, it enters the cooling bed for cooling, and then enters the low-temperature straightening process; if the flatness does not meet the standard after hot straightening, repeat the process of step (1) in the above process (2), recalculate the straightening specification, and use this specification to straighten the next piece of steel

[0129] 3) Low-temperature straightening (0 °C - 100 °C):

[0130] (1) Detect the inlet shape before low-temperature straightening, and use the high-precision defect detection module to obtain the cross-sectional dimensions and flatness data before low-temperature straightening;

[0131] (2) Based on the cross-sectional dimensions and flatness data before straightening, the reduction for this multi-pass straightening is calculated to be 9 / 8 / 6 / 4 / 3 / 2 / 1 / 1 mm using the core algorithm pressure distribution model;

[0132] (3) During the straightening process, the temperatures of the web and flange of the section steel are detected in real time to ensure that the straightening temperature is 65 ± 3 °C. If the temperature does not meet the standard, the cooling step speed is adjusted according to the temperature calculation model;

[0133] (4) Detect the sectional dimension and straightness data of the section steel in real time after straightening. If, after sizing and multiple-length cutting after straightening, the head accuracy of the H-section steel is ≤ 0.5 mm / 1000 mm, the middle accuracy of the H-section steel is ≤ 0.3 mm / 1000 mm, and the calculated value of the residual stress < 50 MPa, then enter the medium-temperature straightening process; if the straightness does not meet the standard after hot straightening, then recalculate the process of the pressure distribution model in the above processes 1), 2) and 3), recalculate the straightening schedule, and use this schedule to straighten the next piece of steel.

[0134] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the purpose and scope of the present technical solution, and they should all be covered by the scope of the claims of the present invention.

Claims

1. An adaptive straightening system for profiled steel based on multi-physical-field coupling, characterized in that: Comprising: A multi-stage temperature control module for performing three-stage temperature control on the section steel, namely high-temperature straightening, medium-temperature straightening, and low-temperature straightening in sequence. The high-temperature straightening temperature is 600°C to 900°C, the medium-temperature straightening temperature is 400°C to 600°C, and the low-temperature straightening temperature is 0°C to 100°C. And the module integrates a temperature sensor and a PID algorithm to dynamically compensate the surface temperature of the straightening rolls, with the control error less than ±5°C; An adaptive pressure distribution module configured to calculate the pressure distribution values at each straightening point based on the function of the elastic modulus of the section steel material varying with temperature, the second moment of the cross-section at each straightening point, the material viscosity dynamic correction coefficient, and the deformation rate influence factor function, and update the pressure parameters every 500 ms; A high-precision defect detection module including a high-density laser array and a deep learning classifier. The laser array generates three-dimensional topography point cloud data of the section steel at a sampling rate of ≥2000 points / second. The classifier is based on the ResNet or MobileNet architecture and uses Focal Loss to optimize defect classification; A dynamic feedback control module for fusing the data of pressure, temperature, and deformation sensors, constructing a closed-loop feedback system with a 500 ms control cycle, and combining the Mixture of Experts (MoE) model to predict the deviation trend of the straightening parameters to achieve dynamic compensation.

2. The self-adaptive straightening system for profiled steel based on multi-physical-field coupling according to claim 1, wherein: In the high-temperature straightening stage, an induction heat compensation device is used to adjust the temperature of the web and flange of the section steel to the set range; in the low-temperature straightening stage, a semiconductor refrigeration technology is used to control the temperature of the straightening rolls, and the cooling step speed is dynamically adjusted based on the temperature model.

3. The adaptive straightening system for profiled steel based on multi-physical-field coupling according to claim 1, wherein: The pressure distribution model of the adaptive pressure distribution module satisfies: P i is the pressure value at the i-th straightening point, E(T) is the elastic modulus temperature function, I i is the second moment of the cross-section, μ is the viscous correction coefficient, ∑W j is the normalized weight coefficient, is the correction factor function including the Poisson's ratio and the deformation rate.

4. The self-adaptive straightening system for profiled steel based on multi-physical-field coupling according to claim 1, characterized in that: The closed-loop feedback system converges the residual stress after straightening to <50 MPa, and controls the head accuracy of the section steel ≤0.5 mm / 1000 mm and the middle accuracy ≤0.3 mm / 1000 mm.

5. The self-adaptive straightening system for profiled steel based on multi-physical-field coupling according to claim 1, wherein: The Mixture of Experts (MoE) model is trained with historical data, can predict the deviation of the straightening parameters 30 seconds in advance, and adjusts the reduction amount and roll speed through a dynamic compensation strategy.

6. The self-adaptive straightening system for profiled steel based on multi-physical-field coupling according to claim 1, wherein: The system is adapted to a cantilever or gantry multi-roll straightening machine, covering H-shaped steel, rails, and high-strength steel with specifications from 100 mm×50 mm to 1750 mm×550 mm, and the outlet speed reaches 10.0 m / s.

7. A prediction and compensation method for section steel straightening based on multi-physical field coupling, characterized in that: Including the following steps: Step 1: Perform high-temperature straightening at 600°C to 900°C, medium-temperature straightening at 400°C to 600°C, and low-temperature straightening at 0°C to 100°C on the section steel in sequence, and compensate the temperature of the straightening rolls in real time through the PID algorithm at each stage, with the error controlled within ±5°C; Step 2: Calculate the pressure distribution values at each straightening point according to the elastic modulus temperature function, the second moment of the cross-section, the viscosity correction coefficient, and the deformation rate of the section steel, and update the pressure parameters through real-time optimization with a 500 ms cycle; Step 3: Scan the surface of the section steel with a high-density laser array to generate three-dimensional topography point cloud data, and identify defects through a deep learning classifier. The classifier uses Focal Loss to optimize multi-class classification; Step 4: Integrate the feedback data from the pressure, temperature, and deformation sensors, construct a 500-ms closed-loop control cycle, combine the Mixture of Experts (MoE) model to predict the deviation trend of the straightening parameters, and dynamically adjust the reduction and roll speed to make the residual stress < 50 MPa.

8. The method for predicting and compensating the straightening of profiled steel based on multi-physical field coupling according to claim 7, wherein: In Step 1, if the flatness after straightening in a certain stage does not meet the standard, recalculate the pressure distribution parameters for this stage and subsequent stages, and apply the corrected parameters to the next section of the section steel.

9. The method for predicting and compensating the straightening of profiled steel based on multi-physical-field coupling according to claim 7, wherein: In Step 3, the deep learning classifier performs image enhancement and sample expansion on the defect data, and is trained based on the ResNet or MobileNet architecture, with a classification response time ≤ 100 ms.

10. The method for predicting and compensating the straightening of profiled steel based on multi-physical field coupling according to claim 7, characterized in that: In Step 4, the Mixture of Experts (MoE) model generates compensation parameters through the real-time collected strain rate, temperature gradient, and historical process data, and the prediction time window is 30 seconds.

Images