Laser cutting method for spliced building steel plates and cutting production line structure

Abstract

The invention relates to the technical field of building steel plate processing, in particular to a laser cutting method and a cutting production line structure for spliced building steel plates, which comprises the following steps of: S1, preliminary rapid detection, S2, uneven area length measurement, S3, three-dimensional data acquisition and S4, cutting link. S2, triggering execution of the step S2 according to a detection result of the step S1 and providing an initial scanning area, providing a precise scanning area for the step S3 according to the unsmooth area range determined in the step S2, and providing a parameter adjustment basis for the step S4 according to the three-dimensional form data obtained in the step S3 so as to realize precise control of the whole process from detection to cutting. The invention aims to solve the potential safety hazard problem that when spliced building steel plates are cut in the prior art, particularly under the condition that joints are uneven, the cutting precision is obviously reduced, and the strength and stability of a building structure are possibly reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of building steel plate processing, and in particular to a laser cutting method and a cutting production line structure for spliced ​​building steel plates. Background Art

[0002] In modern buildings, steel structures are widely used due to their excellent performance, and spliced ​​steel plate structures are crucial in large buildings and infrastructure. Laser cutting technology has become a key means of steel plate processing due to its high precision and efficiency. However, the existing laser cutting technology has limitations when processing and splicing building steel plates, and it is difficult to meet the production needs of high quality and high efficiency. Traditional laser cutting technology is mainly used for single flat steel plates, but in actual production lines, especially when continuously processing and splicing steel plates, it faces the following technical bottlenecks:

[0003] 1. Processing difficulties at the joints: Traditional processes ignore the unevenness and thickness changes at the joints caused by welding and other processes. This makes the cutting parameters set for flat steel plates no longer applicable, resulting in reduced cutting quality and efficiency at the joints. The existing technology has deficiencies in continuous high-quality cutting at the joints, affecting production efficiency and material utilization. 2. Uneven steel plates are common: Building steel plates are easily deformed during production, transportation and installation, resulting in unevenness such as warping and bending. This unevenness makes it difficult to accurately control the relative position of the steel plate surface and the laser cutting head. Traditional fixed-focus laser cutting heads are difficult to adapt to the ups and downs of the steel plate surface. 3. Insufficient fixed focus accuracy: When a fixed-focus laser cutting head faces an uneven steel plate, the focus of the laser beam is likely to deviate from the actual surface of the steel plate and exceed the effective cutting range. This results in the inability of the laser energy to effectively act on the cutting part, reducing cutting efficiency and quality, and even possible cutting failure. Especially in key joints, reduced accuracy may affect the safety of building structures. Existing technologies are difficult to accurately match uneven steel plates, limiting the application of laser cutting.

[0004] In summary, the existing laser cutting technology faces challenges in the processing of spliced ​​building steel plates, such as blank processing at the splicing, poor adaptability to uneven steel plates, insufficient precision, and lack of real-time monitoring. Solving the problem of dynamic matching and precision improvement of laser cutting heads at the splicing of uneven steel plates is a key technical problem that needs to be solved urgently, especially in the context of the increasing safety and quality requirements of the construction industry. Summary of the invention

[0005] In order to solve the technical problems of the above-mentioned prior art to a certain extent as much as possible, the present invention provides a laser cutting method and a cutting production line structure for spliced ​​building steel plates, aiming to solve the safety hazard problem that when cutting and splicing building steel plates, especially when there are unevenness at the joints, the cutting accuracy is significantly reduced, which may lead to a decrease in the strength and stability of the building structure.

[0006] The present invention discloses a laser cutting method for splicing building steel plates. The method first performs step S1, i.e., a "preliminary rapid detection" step. In this step, the system uses a group of multiple laser beams arranged linearly along the length direction of the joint of the spliced ​​steel plate, and projects them to the joint area of ​​the steel plate at a preset tilt angle. The reason for using multiple laser beams and tilted projection is to be able to quickly and sensitively capture the uneven deformation that may exist at the joint. The obliquely projected light will form a regular projection on the flat surface. Once there is a height difference or angle deviation at the joint, the light projection will be dislocated and deformed. The system monitors the deformation of these light projections in real time through image analysis technology. If the analysis result shows that the light projection has a dislocation deformation exceeding the preset threshold, it is determined that there is unevenness at the joint, and the subsequent step S2 is triggered, indicating that the preliminary detection has found an area that needs further precise measurement. On the contrary, if no obvious deformation is detected, it may indicate that the joint is relatively flat, and the subsequent fine measurement step can be skipped (although this embodiment does not clearly indicate the processing method of the flatness, in actual application, it can be considered to directly perform standard cutting or re-verification).

[0007] In step S2, the "uneven area length measurement" step, the system switches the working mode of the laser to a single transverse light mode. Unlike the rapid screening of multiple light rays in step S1, step S2 aims to accurately measure the length range of the uneven area. A single transverse light is projected perpendicular to the length direction of the joint and scans and moves along the length direction of the joint. During the movement, the system continuously monitors the deformation of the single light and records the starting position where the light begins to deform significantly and the ending position where the deformation disappears. These two position points define the range of the uneven area in the length direction of the joint, thereby determining the length of the uneven area. The measurement results of step S2 provide the length range information of the target area for the subsequent accurate scanning of step S3.

[0008] Step S3, the "stereoscopic data acquisition" step, is a key preparation link for achieving precise cutting. After the length range of the uneven area is determined in the previous step, step S3 uses an ultrasonic probe to scan the area to obtain three-dimensional morphological data of the uneven area. The ultrasonic probe can accurately measure the height information of the surface of an object in a non-contact manner. By performing intensive scanning in uneven areas, the system can construct a three-dimensional point cloud model or depth map of the area. These stereoscopic morphological data accurately describe key geometric information such as the bumps and height changes of the uneven area, and provide a data basis for the dynamic cutting parameter adjustment of the subsequent step S4. The stereoscopic morphological data obtained in step S3 is the parameter basis for precise cutting of uneven areas.

[0009] Finally, step S4, the "cutting link" step, is the final execution step of this embodiment. According to the three-dimensional morphological data of the uneven area collected in step S3, the cutting control system adjusts the key parameters of the laser cutting head in real time during the laser cutting process, including height (Z-axis position), focal length and laser power. This dynamic adjustment is based on the three-dimensional morphology of the uneven area. The purpose is to ensure that the focus of the laser beam is always accurately positioned on the actual surface of the steel plate, and dynamically compensate the laser energy according to the surface undulations, so as to achieve precise cutting of the uneven area. The dynamic parameter adjustment of step S4 is the actuator to achieve precise cutting.

[0010] According to a laser cutting method for splicing building steel plates of the present invention, in step S1, the incident angle of the laser light is Satisfy the formula:

[0011]

[0012] in is the estimated uneven height of the steel plate, and d is the width of the joint between the two steel plates. It can be understood that the design of the laser light incident angle θ = 45° + arctan (h1 / 2d) optimizes the geometric relationship between the laser light and the surface of the steel plate. When the light is incident on the uneven surface at this angle, a displacement of Δd = 2×Δh×sin(θ) will be generated. This angle design achieves the best balance between detection sensitivity and signal quality: the larger the angle, the more obvious light displacement will be produced for the same height of unevenness, thereby improving detection sensitivity; but at the same time, too large an angle will lead to signal attenuation and enhanced edge effects. Through this design, the system can automatically calculate the optimal incident angle based on the estimated uneven height h1 of the steel plate and the width d of the joint, significantly improving detection accuracy and sensitivity and reducing missed detection rate.

[0013] The spacing L of the laser array generating the plurality of laser beams satisfies the relationship: L≤ , where W is the minimum uneven length to be detected. It can be understood that the design of the laser array spacing L≤W×cos(θ) / 3 ensures effective coverage of the smallest uneven area. This spacing design ensures that at least 3 laser beams are simultaneously irradiated to the smallest uneven area, forming a multi-point verification mechanism, which significantly reduces the misjudgment rate caused by single-point light interference. This redundant design enables the system to distinguish between real structural unevenness and interference factors such as surface contamination, improves detection reliability, and ensures the quality stability of the subsequent cutting process.

[0014] The misalignment and deformation of the light projection is determined by a dynamic threshold. When the offset Δd of the light relative to the ideal position meets the condition: Δd>max(0.15mm,0.02×h), it is determined as an uneven area, where h is the thickness of the steel plate. It can be understood that the dynamic threshold Δd > max(0.15mm, 0.02×h) is used to determine the misalignment and deformation of the light, which solves the problem of inconsistent detection standards for steel plates of different thicknesses. The basic threshold of 0.15mm ensures the system's ability to detect tiny unevenness, while the dynamic threshold of 0.02×h takes into account the thickness of the steel plate, because the manufacturing tolerance of thicker steel plates is usually larger, and the detection threshold needs to be appropriately increased. This threshold relationship can reduce the system's misjudgment rate, while ensuring that the detection rate of unevenness above 0.2mm is improved, meeting the quality control requirements for cutting of building steel plates, and laying a high-quality detection foundation for the entire cutting process.

[0015] According to a laser cutting method for spliced ​​building steel plates of the present invention, in step S2, when the laser is converted from a wide-area multi-light mode to a single transverse light mode, a micro-rotating grating is used to converge multiple light sources into a single transverse light, and the same incident angle as step S1 is maintained; the moving speed v of the laser along the length direction of the seam satisfies the relationship: , where D is the length of the steel plate joint to be measured, and T is the maximum measurement time allowed by the system; the deformation judgment adopts the boundary confirmation algorithm, and records the conditions that meet the following conditions: N consecutive frames (N≥3) of images are used as the boundary points of the uneven area, where: is the deformation measurement parameter, is the deformation measurement parameter benchmark value of the ideal plane, and h is the thickness of the steel plate. It can be understood that the above scheme involves the working mode conversion, movement speed control and deformation boundary judgment method of the laser in step S2. The technical effects produced by these features are as follows: the technical feature of the laser converting from a wide-area multi-light mode to a single transverse light mode realizes a seamless switch from "wide-area rapid screening" to "precise directional measurement". The multi-path light source is converged into a single transverse light by a micro-rotating grating, and the same incident angle as step S1 is maintained. This design ensures the consistency and continuity of the detection results, the conversion time is short, and the mode switching time is reduced. This efficient optical path reconstruction method avoids the inconvenience of traditional equipment replacement or long-term adjustment, and improves the system response speed and detection efficiency. In addition, the movement speed of the laser along the length of the seam is v = min[50mm / s, ] design, achieving a dynamic balance between detection accuracy and efficiency. This relationship enables the system to automatically adjust the optimal scanning speed according to the seam length D and the allowed measurement time T. When the boundary of the uneven area presents a gradual change, the scanning speed can be reduced to improve the boundary positioning accuracy, while in the area with clear boundaries, the speed can be increased to improve the detection efficiency. This dynamic speed control strategy reduces the overall detection time and is crucial for continuous operation of the production line. Finally, the deformation judgment adopts the boundary confirmation algorithm, and records satisfying |η-η0| > max(0.15, 0.05× ) continuous N frames (N ≥ 3) of images under the condition of ) are used as the boundary points of the uneven area. This design significantly improves the accuracy of boundary recognition and anti-interference ability. Introducing the square root function (0.05× ) rather than a nonlinear function, which better adapts to the nonlinear relationship between steel plate thickness and manufacturing accuracy. The judgment requirement of N consecutive frames effectively filters out environmental interference and instantaneous misjudgment, allowing the system to maintain a high boundary recognition accuracy on steel plates of various thicknesses in the range of 5-50mm, while significantly reducing the misjudgment rate. This technical effect provides a precise target area for subsequent stereo data acquisition and improves the measurement accuracy of the length of uneven areas.

[0016] According to a laser cutting method for spliced ​​building steel plates of the present invention, in step S3, the starting scanning position P0 of the ultrasonic probe satisfies the relationship: P0=S1-max(20mm, 0.05L), wherein S1 is the starting coordinate of the uneven boundary, and L is the total length of the uneven area; the ultrasonic probe is a phased array ultrasonic probe, comprising a rectangular array of 64-128 piezoelectric transducer units, and adopts a three-band alternating working mode of low frequency band (3-4MHz), medium frequency band (5-7MHz) and high frequency band (8-10MHz), and performs multi-angle focusing scanning of the central vertical beam and ±15°, ±30° deflection beams for each scanning point. It can be understood that the starting scanning position of the ultrasonic probe in step S3 is determined and the configuration and working mode of the phased array ultrasonic probe are determined. The technical effects of these features are as follows: 1. The design of the starting scanning position P0=S1-max (20mm, 0.05L) of the ultrasonic probe ensures the integrity of the boundary area data by reserving an appropriate scanning margin (at least 20mm beyond the uneven area or 5% of its total length), effectively avoiding the subsequent cutting problem that may be caused by the lack of boundary data. This intelligent buffer design has adaptive characteristics, automatically increasing the buffer distance for long uneven areas, and maintaining a sufficient minimum buffer distance for short uneven areas, ensuring that complete transition area data can be obtained in all cases, improving the reliability and integrity of the measurement. The phased array ultrasonic probe uses a rectangular array containing 64-128 piezoelectric transducers to achieve high spatial resolution regional scanning capabilities. The array coverage area (60mm×30mm) is larger than the typical joint width of the steel plate, ensuring that a single scan can completely cover the joint area. The scanning accuracy reaches ±0.05mm in the depth direction and ±0.2mm in the plane direction, which is far superior to traditional single-frequency ultrasonic testing technology. The three-band alternating working mode (low frequency 3-4MHz, medium frequency 5-7MHz, high frequency 8-10MHz) makes full use of the complementary characteristics of ultrasound at different frequencies: low frequency provides deep penetration, high frequency provides fine surface details, and medium frequency balances the two. This design enables the system to obtain comprehensive three-dimensional information in a single scan, and the detection capability is improved compared with single-frequency technology. For steel plates of different thicknesses, the system can dynamically adjust the working frequency. For example, when scanning a 12mm thick steel plate, the center frequency is automatically reduced to 3-5MHz to ensure sufficient penetration depth; while for areas with subtle surface changes, it is increased to 8-10MHz to capture fine details. The multi-angle focused beam scanning (central vertical beam and ±15°, ±30° deflection beam) solves the "blind spot" problem of traditional single vertical incidence ultrasonic detection when facing inclined surfaces. For surfaces with an inclined angle of α, the best detection effect can be obtained when the detection beam angle θ satisfies |θ-α| < 15°.This design ensures that effective echo signals can be obtained for any surface with an inclination angle within the range of ±45°, and the detection integrity is improved by about 25%, especially the detection capability of high-inclination surfaces is significantly enhanced. This technical effect directly improves the comprehensiveness and accuracy of three-dimensional data collection, and provides a reliable data basis for subsequent cutting parameter optimization.

[0017] According to a laser cutting method for splicing building steel plates of the present invention, in step S3, the scanning density D s Adopting adaptive strategy, satisfying the relationship: D s =min[3mm, max(0.5mm, , where G is the deformation gradient of the local area; the uneven area boundary determination adopts gradient analysis technology, when the height gradient G meets the condition: |G|>max(0.1mm / mm, 0.02× / mm), it is determined as a boundary point, where h is the thickness of the steel plate; the system constructs a three-dimensional grid model through multi-source data fusion technology, the grid density is adaptively matched with the surface complexity, the grid is encrypted (0.2-0.5mm) in the complex area of ​​concave and convex changes, and the grid is sparse (1-2mm) in the flat area. It can be understood that the scanning density D s Adopting adaptive strategy, satisfying the relationship: D s =min[3mm, max(0.5mm, )], where G is the deformation gradient of the local area. This design is based on sampling theory (to accurately reconstruct a surface containing a characteristic size λ, the sampling spacing d must satisfy: d≤ ), automatically adjusts the scanning point density according to the geometric complexity of the surface, increases the sampling density in areas with drastic changes (high gradient), and uses lower density scanning in flat areas. This intelligent resource allocation mechanism greatly reduces the scanning time and data storage requirements while ensuring data quality, while improving the accuracy and efficiency of 3D reconstruction. In addition, the gradient analysis technology is used to determine the boundary of the uneven area. When the height gradient G meets the conditions: |G|>max(0.1mm / mm, 0.02× / mm), it is determined as a boundary point. This design achieves accurate identification of the boundary between the uneven area and the normal area by analyzing the spatial gradient of the height data. The positioning accuracy is much higher than the preliminary boundary measurement in step S2. The dynamic threshold related to the thickness of the steel plate (0.02× / mm), so that the system can automatically adjust the boundary judgment standard for steel plates of different thicknesses, and maintain a high boundary recognition accuracy rate for steel plates of various thicknesses in the range of 5-50mm, while reducing the misjudgment rate. This precise boundary recognition technology provides an accurate regional division basis for subsequent cutting parameter adjustment. In addition, multi-source data fusion technology constructs a three-dimensional grid model, and the grid density is adaptively matched with the surface complexity to achieve efficient data representation. The grid encryption (0.2-0.5mm) in the complex area of ​​concave and convex changes ensures the high-precision expression of the key feature area; the grid sparseness (1-2mm) in the flat area significantly reduces the overall complexity of the model. This adaptive grid density design enables the system to significantly reduce the amount of data while maintaining the accuracy of the model, which can reduce more storage space on average, while improving data processing and transmission efficiency. The generated optimized three-dimensional model not only retains the accurate expression of key geometric features, but also greatly reduces the burden of data processing, providing an ideal data basis for subsequent cutting path planning and parameter optimization, which directly affects the final cutting quality and efficiency.

[0018] According to a laser cutting method for splicing building steel plates of the present invention, in step S4, the focus control system adopts a two-stage focusing mechanism, including coarse adjustment of the cutting head height by an electric lifting mechanism and fine adjustment of the focus position by a variable focal length optical system, and the two-stage system works together to satisfy the relationship: ΔZ+ΔF=ΔH+C, wherein ΔZ is the height adjustment amount, ΔF is the focal length adjustment amount, ΔH is the surface height change amount, and C is a constant correction term; the laser power dynamic adjustment of the laser cutting head satisfies the relationship: P=P0×[1+k× ], where P is the adjusted power, P0 is the standard power, k is the material-related coefficient, ΔF is the focal length change, and F0 is the reference focal length. It can be understood that the focus control system adopts a two-stage focusing mechanism, including coarse adjustment of the cutting head height through the electric lifting mechanism and fine adjustment of the focus position through the variable focal length optical system. The two-stage system works together to satisfy the relationship: ΔZ+ΔF=ΔH+C. This design achieves full-range precision focus control: the coarse adjustment mechanism provides a wide range of adjustment capabilities (50-100mm) to adapt to uneven height changes of various scales; the fine adjustment mechanism provides high-frequency (50Hz) and high-precision (±2mm) fine adjustment capabilities to ensure precise focus positioning. When the two work together, the system can quickly respond to surface height changes ΔH, and achieve the best dynamic response characteristics by reasonably allocating mechanical height adjustment ΔZ and optical focal length adjustment ΔF. The constant correction term C (usually ±0.05mm) provides a system error compensation function, offsets the inherent deviations of the mechanical system and the optical system, and improves the accuracy and stability of the overall system. In addition, the two-stage focusing mechanism enables the system to adapt to uneven areas with a height difference of ±5mm, and the dynamic adjustment accuracy reaches ±0.05mm, which solves the problem of the traditional fixed focus system causing the focus position to deviate when cutting in uneven areas, resulting in a decrease in cutting energy density. The cutting accuracy can also be significantly improved, and the consistency of cutting quality is significantly improved. In addition, the dynamic adjustment of the laser power of the laser cutting head satisfies the relationship: P=P0×[1+k×|ΔF|² / (F0)²]. This design is based on the optical physics characteristics of the laser. It accurately calculates the impact of focal length changes on the focal size (focal size A∝(1+|ΔF / F0|)²) and adjusts the laser power accordingly to ensure that the energy density at the focus remains constant during the focal length adjustment process, achieving consistent cutting results. The introduction of the material correlation coefficient k (usually 0.15-0.25 for steel) enables the system to automatically adjust the power compensation coefficient according to different material properties, improving the system's adaptability to different types of steel. This dynamic power adjustment mechanism solves the problem of unstable cutting quality when the focal length changes in traditional fixed power systems, significantly improves the consistency of cutting depth, improves the stability of surface quality, and improves energy efficiency. The system can achieve a cutting accuracy of ±0.1mm in uneven areas, and even in areas with a height difference of ±5mm, the slit width change is still controlled within the range of ±0.05mm, providing key technical guarantees for the manufacture of high-quality building steel structures.

[0019] According to a laser cutting method for spliced ​​building steel plates of the present invention, in step S4, a feedforward control strategy is adopted in a high-speed cutting scenario, the surface morphology of the next 0.1-0.2 seconds is predicted based on the three-dimensional model of the uneven area, the control instructions are calculated and cached in advance, and the time window is predicted. Satisfies the relationship: ,in, To control the delay, For system execution delay, is the safety factor; the feedforward control is when the cutting speed v exceeds the critical value Automatically activated when the threshold Set to 30mm / s. It can be understood that in the high-speed cutting scenario, a feedforward control strategy is adopted to predict the surface morphology of the next 0.1-0.2 seconds based on the three-dimensional model of the uneven area, and the control instructions are calculated and cached in advance. This design solves the "tracking error" problem caused by the traditional feedback control system during high-speed cutting, which is caused by the actual adjustment action lagging behind the surface change due to control delay. During high-speed cutting, the tracking error Δe satisfies the relationship with the cutting speed v, control delay τ and surface gradient G: Δe = v × τ× G. Feedforward control effectively compensates for this delay by predicting the future surface morphology and performing adjustment actions in advance, so that the control action is accurately synchronized with the actual demand, and the dynamic adjustment error is controlled within ±0.1mm, which is about 70% higher than the dynamic accuracy of traditional feedback control. Prediction time window Satisfies the relationship: , where τ is the control delay, For system execution delay, This adaptive prediction range design enables the system to automatically adjust the prediction time window according to the actual delay situation, ensuring that there is always enough time to prepare control actions in advance and adapt to different system configurations and operating conditions. Safety factor (usually 1.5-2.0) provides additional time margin, enhances the system's adaptability to delay fluctuations, and maintains effective feedforward control even when system load changes or temporary interference factors occur. The design of automatic activation when the speed reaches 30mm / s achieves the optimal allocation of control resources. Ordinary feedback control can meet the accuracy requirements when cutting at low speeds, and the feedforward control with large calculation amount is activated only when cutting at high speeds, avoiding unnecessary complex calculations. The clear speed critical value setting enables the system to automatically select the most suitable control strategy according to the cutting speed, achieving a smooth transition from feedback control to feedforward control, and avoiding the instability that may be caused by switching control strategies.

[0020] According to a laser cutting method for spliced ​​building steel plates of the present invention, in step S4, the cutting process is monitored in real time by photoelectric sensors, infrared temperature sensors, acoustic sensors and reflected light intensity sensors, and a comprehensive cutting quality index is defined: Q=w1C+w2T+w3A+w4R, wherein C is the consistency score of the slit width, T is the rationality score of the temperature distribution, A is the regularity score of the acoustic characteristics, R is the stability score of the reflected light intensity, and w1 to w4 are weight coefficients; when Q is lower than a threshold value, Trigger parameter fine-tuning, =0.85-0.05×( ), where G is the local area unevenness measure, G0 is the reference value; the parameter fine-tuning amplitude satisfies the condition: ≤0.15, ≤0.2, ≤0.3mm, where ΔP is the power adjustment, ΔV is the speed adjustment, and ΔZ is the height adjustment. It can be understood that the cutting process is monitored in real time by photoelectric sensors, infrared temperature sensors, acoustic sensors, and reflected light intensity sensors, and the comprehensive cutting quality index Q=w1C+w2T+w3A+w4R is defined. This multi-sensor fusion monitoring system realizes a comprehensive evaluation of cutting quality: photoelectric sensors monitor the slit width and consistency, infrared temperature sensors monitor the temperature distribution in the cutting area, acoustic sensors monitor the cutting noise spectrum characteristics, and reflected light intensity sensors monitor the focus position deviation. Multiple sensors work together to form a comprehensive monitoring of cutting quality, which can capture complex quality problems that are difficult for a single sensor to identify, and significantly improve the accuracy and comprehensiveness of quality assessment. Different sensors have different sensitivities to various types of cutting anomalies. Multi-sensor fusion enables the system to detect potential anomalies early before the problem seriously affects the cutting quality. Compared with single sensor monitoring, the average advance amount of anomaly detection is significantly improved. Cutting quality threshold =0.85-0.05×( ) design, realizing adaptive quality standards. The basic threshold of 0.85 represents the minimum cutting quality standard required by the system in normal flat areas. The dynamic adjustment item -0.05×(G / G0) introduces the degree of unevenness G as a variable, setting reasonable quality expectations for areas with different degrees of unevenness. The relatively flat areas maintain higher standards (close to 0.85), and the standards are appropriately relaxed in highly uneven areas, but the basic quality requirements are still ensured, making quality control both strict and realistic. This dynamic threshold mechanism avoids the problem of over-adjustment in difficult cutting areas, reduces the number of unnecessary parameter adjustments, improves system stability by about 35%, and ensures timely intervention when necessary to maintain quality balance. In addition, the parameter fine-tuning amplitude meets the conditions: |ΔP / P0|≤0.15, |ΔV / V0|≤0.2, |ΔZ|≤0.3mm. By setting the maximum amplitude limit of each parameter adjustment, the system oscillation or loss of control caused by over-adjustment is avoided, ensuring that even in extreme cases, the adjustment action remains within a safe control range. Limiting the range of a single adjustment prompts the system to improve the cutting quality in small steps rather than large changes. This method gradually optimizes the parameters while maintaining the continuity of the cutting, and ultimately achieves a better overall effect. Compared with large adjustments, the resulting cutting marks have a smoother transition. At the same time, by limiting the adjustment range of each parameter and considering the combined influence of multiple parameters, the system can find the best combination of power, speed and height. This collaborative optimization improves the effectiveness of parameter adjustment. In summary, these technical features jointly realize intelligent monitoring and precise adjustment of the cutting process, enabling the system to maintain high-quality cutting in complex and uneven areas. Compared with traditional open-loop control, the cutting accuracy of this system is significantly improved and the scrap rate is significantly reduced. The combination of closed-loop real-time monitoring and adaptive parameter adjustment enables the system to cope with various complex cutting scenarios, including changes in steel plate thickness, uneven materials, etc., to ensure the stability and consistency of cutting quality, and provide technical guarantees for the safety and reliability of building steel structures.

[0021] According to a laser cutting method for spliced ​​building steel plates of the present invention, in step S1, an industrial-grade high-speed CCD camera system with a resolution of not less than 2048×1536 pixels is used, and a narrow-band filter with a central wavelength of 650nm±15nm is equipped to form a three-dimensional monitoring system; in step S2, the frame rate of the camera system is increased to 120-240 frames / second, and the real-time image processing unit adopts a parallel computing architecture, and the processing delay is controlled within 5 milliseconds; between step S3 and step S4, a dual-channel redundant mechanism is used to transmit three-dimensional morphological data, the main channel adopts Gigabit Ethernet direct connection, and the backup channel adopts a dedicated serial interface to ensure transmission reliability.

[0022] The present invention also discloses a cutting production line structure, which is used to implement the laser cutting method of spliced ​​building steel plates of the present invention. The structure includes: a preliminary detection module, including a laser irradiation unit and a preliminary deformation analysis unit, which are used to irradiate the joint area and analyze the dislocation deformation, determine whether there is unevenness, and output an unevenness trigger signal; a length measurement module, including a laser scanning unit and a length determination unit, which scans the joint area after receiving the unevenness trigger signal, records the starting and ending positions of the deformation, and outputs the length data of the uneven area; a three-dimensional data acquisition module, including an ultrasonic scanning unit and a morphological data construction unit, receives the uneven area range data, scans and constructs the three-dimensional morphological data of the uneven area; a dynamic cutting module, including a laser cutting unit and a cutting parameter adjustment unit, adjusts the height, focal length and power of the laser cutting head according to the three-dimensional morphological data, and realizes precise cutting of the uneven area; wherein the detection result of the preliminary detection module triggers the length measurement module, determines the uneven area range and provides a precise scanning area for the three-dimensional data acquisition module, and the collected data provides a real-time cutting parameter adjustment basis for the dynamic cutting module, thereby realizing precise cutting control.

[0023] The technical effects that can be achieved by the laser cutting method of spliced ​​building steel plates of the present invention are as follows: 1. Improving cutting accuracy: Through the precise detection and stereoscopic data acquisition system of four steps (preliminary rapid detection, measurement of the length of the uneven area, three-dimensional data acquisition and cutting link), the morphological information of the uneven area at the joint is accurately obtained, and the laser cutting head parameters are dynamically adjusted according to this information to ensure that the laser beam always acts accurately on the surface of the steel plate, significantly improving the cutting accuracy of the uneven area. 2. Enhancing the safety of the steel plate structure: high-precision cutting and high-quality seam cutting, especially in the key splicing parts of the building steel plate, can ensure the strength and stability of the splicing structure, avoid stress concentration caused by cutting errors or defects, reduce the risk of structural damage, and improve the overall safety of the building structure. 3. Improving production efficiency: By combining preliminary rapid detection and precise measurement, only the uneven area is finely processed, and the efficient conventional cutting mode can still be used for the flat area. The dynamic adjustment process is automated, which reduces manual intervention and improves overall production efficiency. 4. Improve material utilization: Precise cutting reduces scrap and rework, especially when cutting complex contours or special-shaped splicing parts, maximizes the use of steel plate materials, reduces material waste, improves material utilization, and reduces production costs. These technical effects jointly solve the problem of low cutting accuracy in the existing technology when cutting and splicing building steel plates, especially when there are uneven joints, thereby avoiding potential safety hazards that may reduce the strength and stability of the building structure. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a flow chart of the steps of the present invention. DETAILED DESCRIPTION

[0025] In order to more clearly illustrate the technical solution of the present invention, the present invention will be described in detail below in conjunction with specific embodiments. It should be noted that the following embodiments are only used to explain the present invention, rather than to limit the scope of protection of the present invention.

[0026] This embodiment describes a laser cutting method for spliced ​​building steel plates, which is particularly suitable for solving the high-precision laser cutting requirements when there is unevenness at the joints of spliced ​​building steel plates. The method mainly includes the following steps: step S1, preliminary rapid detection; step S2, uneven area length measurement; step S3, three-dimensional data collection; step S4, cutting step.

[0027] Step S1, preliminary rapid detection. Step S1 is intended to conduct a preliminary rapid detection of the joint area, determine whether there is unevenness, and provide basic information for subsequent precise measurement and cutting. The core of this step is to use a plurality of laser light arrays to project to the joint area of ​​the spliced ​​steel plates at an inclined angle, and capture the deformation of the light projection through a high-speed industrial camera system, so as to quickly determine whether there is unevenness at the joint. In this embodiment, the specific implementation process of step S1 is as follows: Step 1.1, precise conveying and positioning of steel plates. First, the spliced ​​steel plates to be cut are conveyed to the inspection station using a high-precision conveying system. The conveying system uses multiple sets of anti-slip coated synchronous conveying rollers driven by servo motors to ensure that the steel plates are conveyed forward along the production line at a stable speed of 2-5 meters / minute. In order to achieve precise positioning, infrared sensor arrays are set on both sides of the production line to monitor the position of the steel plates in real time. When the infrared sensor detects that the leading edge of the steel plate enters the preset area, the deceleration mechanism is triggered to reduce the conveying speed to 0.5-1 meters / minute. Furthermore, a precision electronic rangefinder is set 2 meters upstream of the detection area to continuously monitor the moving distance of the steel plate. When the distance meter determines that the front edge of the steel plate reaches the preset coordinates of the front edge of the detection position, the control system sends a stop signal, the conveyor roller stops rotating, and the initial longitudinal positioning of the steel plate is completed. After the steel plate stops, the four-corner hydraulic positioner is started to apply uniform clamping force from the four corners of the steel plate to ensure that the relative position error of the steel plate on the horizontal plane does not exceed ±0.5mm, providing a stable platform for subsequent high-precision detection. The servo response speed of the above positioning process does not exceed 50 milliseconds, which effectively avoids the impact on the overall efficiency of the production line.

[0028] Through the above-mentioned precise conveying and positioning mechanism, the spliced ​​steel plates can be quickly and accurately positioned in the inspection area, providing a precise initial position for subsequent laser inspection, thereby ensuring the accuracy and consistency of the inspection results.

[0029] Step 1.2, laser system configuration and light projection. After completing the positioning of the steel plate, start the laser system for light projection. This embodiment adopts a multi-channel laser array, and 20 semiconductor laser transmitters are installed at equal intervals along the length direction of the steel plate joint. Each laser transmitter outputs visible red light with a wavelength of 650nm±10nm, and the diameter of a single spot is 1.2mm. The laser array is installed as a whole on a precision bracket that can rotate 360°, and the incident angle θ of the laser light relative to the surface of the steel plate is adjusted by a precision angle driver. According to the technical solution of the present invention, the incident angle θ of the laser light needs to be set according to the estimated uneven height h of the steel plate and the width d of the joint between the two steel plates to ensure the sensitivity of the detection. In this embodiment, the estimated maximum uneven height h of the steel plate is 5mm, the width d of the joint is 1mm, and according to the formula:

[0030]

[0031] The calculated incident angle θ ≈ 68.2°. Considering the actual application scenario, the incident angle θ is set to 60°, which can not only ensure the detection sensitivity but also avoid signal attenuation caused by excessive light incident angle.

[0032] The spacing L of the laser array also needs to be reasonably set to ensure that the light density is sufficient to detect the smallest uneven area. The spacing L of the laser array that generates the multiple laser lights satisfies the relationship:

[0033] L≤

[0034] Among them, the minimum uneven length W that needs to be detected is determined to be 50mm according to the building standard, the incident angle θ is 60°, and the maximum spacing of the laser array is calculated to be L ≈ 8.3mm. To ensure the reliability of detection, this embodiment sets the spacing L of the laser array to 8mm to ensure that each laser light can cross the joint of two adjacent steel plates, and the light density is sufficient to detect the smallest size uneven area. The power of the laser light source is controlled at 4mW, which can ensure that the light projection is clear and visible without causing safety hazards to the steel plate surface or operators. Through the above-mentioned laser system configuration and light projection, a uniform laser light array with a certain incident angle can be formed in the seam area of ​​the spliced ​​steel plates, providing the necessary optical signals for subsequent image acquisition and unevenness identification.

[0035] Step 1.3, camera monitoring system. In order to capture the changes of laser light projection at the joint of steel plates in real time, this embodiment adopts a high-speed industrial camera system. Two industrial-grade high-speed CCD cameras are installed 2.5 meters above the steel plate. The two cameras are arranged at an angle of 15° to form a three-dimensional monitoring system, and the field of view covers the entire joint area. The camera resolution is 2048×1536 pixels and the frame rate is 100 frames per second, ensuring that tiny light deformations can be captured. The camera lens has a focal length of 25mm, an aperture of F2.8, and is equipped with a narrow-band filter with a central wavelength of 650nm±15nm, which effectively enhances the recognition rate of laser light and filters out ambient stray light interference. The camera system adopts a hard triggering method to achieve precise synchronization with the laser, and the trigger delay is less than 5 milliseconds, ensuring the synchronization of image acquisition and light projection. The images collected by the camera system are transmitted to the image processing unit in real time via Gigabit Ethernet, and the transmission delay is less than 8 milliseconds, meeting the needs of real-time processing. The above-mentioned high-speed industrial camera system can achieve high-precision stereoscopic imaging of the steel plate joint area, providing a reliable data basis for subsequent image analysis and unevenness identification.

[0036] Step 1.4, Identification and determination of light misalignment. After receiving the image transmitted by the camera system, the image processing unit uses a specially developed image processing algorithm to identify and determine light misalignment. First, the image processing unit extracts the projection position of each laser light on the steel plate surface and establishes a baseline reference model, which represents the projection position of the laser light on the steel plate surface under an ideal flat state. Subsequently, the system performs real-time analysis on the shape and position of each laser light projection and detects its offset Δd relative to the baseline reference model.

[0037] According to the dynamic threshold determination method of the present invention, when the detected offset Δd of the light relative to the ideal position satisfies the following conditions, it is determined to be an uneven area:

[0038] Δd>max(0.15mm,0.02×h)

[0039] Wherein, h is the nominal thickness of the steel plate. In this embodiment, the thickness of the steel plate is 10 mm, and the judgment threshold is calculated to be Δd > max(0.15 mm, 0.2 mm) = 0.2 mm. That is, when the light offset exceeds 0.2 mm, the system determines that there is unevenness in the area. The time response of the system to determine unevenness does not exceed 80 milliseconds, ensuring the real-time detection. For areas determined to be uneven, the system automatically records the corresponding laser number and its position coordinates on the joint, providing initial information for the precise measurement of the subsequent step S2.

[0040] Step S2, measuring the length of the uneven area. Step S2 is started after the unevenness at the joint is initially detected in step S1. Its purpose is to accurately measure the size of the uneven area in the length direction of the joint, provide an accurate scanning range for the subsequent three-dimensional data acquisition in step S3, and provide a length basis for the final dynamic cutting parameter adjustment. The key to this step lies in the switching of the laser working mode, high-precision scanning motion control, real-time image acquisition and processing, and accurate identification of the boundary of the uneven area. In this embodiment, the specific implementation process of step S2 is described in detail as follows:

[0041] Step 2.1 Working mode conversion mechanism. After determining that there is unevenness at the joint in step S1, the control system immediately starts the working mode conversion program and switches the laser system from the wide-area multi-light rapid detection mode to the single transverse light precise measurement mode. The purpose of this conversion process is to focus the laser light source for more refined scanning measurements. Specifically, the control unit sends a mode switching signal to the laser array, and the system's built-in multi-way selection switch completes the state switching within 3 milliseconds. The optical path switching device in the laser assembly is then started, and the originally multi-path divergent laser beams are converged into a single transverse light using a micro-rotating grating. The diameter of this single transverse light is precisely adjusted to 1.0 mm to improve the measurement resolution. Importantly, during the mode switching process, the single transverse light maintains the same oblique incident angle θ (60° in this embodiment) as the laser light in step S1. To ensure the accuracy of the angle, the system starts a precision calibration program and verifies the light projection angle using a preset reference calibration block. When the angle error exceeds ±0.3°, the automatic fine-tuning mechanism is triggered to precisely adjust the angle of the rotating grating until the angle error returns to the allowable range. After the mode switch is completed, in order to improve the signal-to-noise ratio of a single light beam, the laser emission power is automatically increased to 9mW, but it is still controlled within the power range that is safe for the steel plate surface and the operator. The entire mode conversion process is completed within 180 milliseconds, achieving rapid and seamless switching of detection modes, ensuring the continuity and efficiency of the detection process. Through the above-mentioned working mode conversion mechanism, the laser system can quickly switch from wide-area detection to fine measurement mode, providing the necessary light source conditions for subsequent uneven area length measurements.

[0042] Step 2.2 High-precision motion control system. In order to achieve accurate scanning of a single horizontal laser beam along the length direction of the seam, this embodiment adopts a high-precision motion control system. The core component of the system is a retractable device, which integrates the main telescopic mechanism and the precision displacement platform to jointly achieve accurate positioning and smooth movement of the laser. The main telescopic mechanism adopts a synchronous toothed belt drive driven by a servo motor, which is responsible for the coarse positioning of the laser in the length direction of the seam, and its positioning accuracy reaches ±0.8mm. The precision displacement platform is driven by a linear motor and has sub-micron positioning capability. The repeatability positioning accuracy is better than ±0.008mm, and it is responsible for fine displacement control during the scanning process. Among them, the scanning speed v of the laser along the length direction of the seam adopts a dynamic adjustment strategy, which satisfies the relationship:

[0043]

[0044] Wherein, D is the length of the steel plate joint to be measured (estimated by the preliminary detection information in step S1, or input by the operator), and T is the maximum measurement time allowed by the system, which is set to 20 seconds in this embodiment. For example, when the joint length D is 1000mm, the calculated scanning speed v = min[50mm / s, (1000mm / (10 * 20s))] = 5mm / s.

[0045] The motion control system adopts a three-stage speed curve, including a smooth acceleration stage, a constant speed scanning stage and a smooth deceleration stage. The acceleration and deceleration are controlled at 0.4g to reduce inertial shock and vibration. In order to further eliminate the influence of mechanical vibration on measurement accuracy, the guide rail system is equipped with an active vibration reduction device, which uses piezoelectric ceramic actuators to offset vibration in real time. When the amplitude exceeds 0.02mm, the vibration reduction device starts immediately. The high-resolution position encoder feeds back the laser position information in real time, with a resolution of 0.003mm, ensuring that the system accurately grasps the spatial coordinates of each measuring point. Through the above-mentioned high-precision motion control system, a single transverse laser beam can be stably and accurately scanned along the length of the seam, providing a reliable motion platform for subsequent image acquisition and identification of uneven area boundaries.

[0046] Step 2.3 Real-time image acquisition and processing. During the laser scanning process, the camera monitoring system synchronously enters the high-speed acquisition mode, the frame rate is increased to 200 frames per second, and the exposure time is shortened to 1.5 milliseconds to avoid image blur caused by the movement of the laser. The camera field of view is adjusted to the narrowband focus mode, and the field of view width W is adaptively adjusted according to the range of the uneven area detected in step S1, and satisfies the following relationship: W = 2S + 30mm, where S is the distance between the farthest adjacent uneven points detected in step S1. In addition, the image acquisition system adopts polling cache technology and sets a three-level cache queue to ensure that no frame of image is lost during high-speed acquisition. The real-time image processing unit performs the following processing flow: S1, narrowband filtering: apply the narrowband filtering algorithm to further filter out the ambient stray light and accurately extract the contour of the laser light; S2, curvature and offset calculation: calculate the actual projection curvature K and position offset D of the laser light at each acquisition point. The curvature reflects the degree of light deformation, and the offset reflects the height of the unevenness; S3, deformation measurement parameter generation: compared with the ideal light model (straight line), the deformation measurement parameter η = f(K, D) is generated. This parameter comprehensively reflects the degree of light deformation. The larger the value, the more serious the deformation; S4, data association storage: the deformation measurement parameter η is associated with the timestamp and spatial position information to form an orderly measurement data stream. In order to meet the real-time requirements, the image processing unit adopts a parallel computing architecture, processes the image in blocks, and controls the processing delay to less than 4 milliseconds. The above real-time image acquisition and processing mechanism can capture the subtle changes of laser light projection at high speed and precision, and convert the light deformation information into quantifiable deformation measurement parameters, providing key data for subsequent uneven area boundary identification and length calculation.

[0047] Step 2.4 Boundary recognition and length calculation. This embodiment uses an accurate boundary recognition algorithm to determine the actual boundary of the uneven area based on the deformation measurement parameter η output by the real-time image processing unit. The system analyzes the deformation measurement parameter η point by point and determines the boundary judgment condition:

[0048]

[0049] Determine whether it is a boundary point of the uneven area. Among them, η0 is the deformation measurement parameter reference value of the ideal plane (measured in the flat area), and h is the thickness of the steel plate. In this embodiment, the thickness of the steel plate h is 10mm, the reference value η0 is calibrated to 0.05, and the boundary judgment threshold is calculated to be max(0.15, 0.05× ) ≈ 0.158. That is, when the deformation measurement parameter η deviates from the reference value η0 by more than 0.158, it is determined to be a boundary point of the uneven area. In order to prevent the system from misjudging instantaneous interference, boundary confirmation must meet the time continuity condition: the deformation measurement parameters in N=4 consecutive frames of images must meet the above conditions before they are finally confirmed as boundary points.

[0050] The system records the starting coordinates S1 and the ending coordinates S2 of the uneven boundary, and calculates the actual length L of the uneven area according to the following formula: L = |S2 - S1| + 2δ, where δ is a safety margin, which is set to 4mm in this embodiment to ensure complete coverage of the uneven area. For uneven areas of complex shapes, the system can identify multiple continuous sub-areas, and the length Li of each sub-area must meet the minimum effective uneven area length Lmin requirement. In this embodiment, Lmin is set to 8mm. After completing the boundary recognition, the system generates an accurate spatial positioning report of the uneven area, including: 1. The coordinate range of the uneven area (starting point S1 and end point S2); 2. The total length L of the uneven area; 3. The severity classification (based on the maximum value of the deformation measurement parameter η, for example, divided into three levels: mild, medium, and severe); 4. The spatial distribution curve of the deformation parameters in the area. This report will serve as the basis for the scanning range of the three-dimensional data acquisition in step S3, and provide length parameters for subsequent cutting path planning.

[0051] Step S3, three-dimensional data acquisition. Step S3 is initiated based on the precise measurement of the length of the uneven area in step S2, and is intended to collect three-dimensional morphological data of the uneven area with high precision, and provide accurate spatial geometric information for the dynamic laser cutting head parameter adjustment in the subsequent step S4. The core of this step lies in the precise positioning of the ultrasonic probe, multi-spectrum phased array ultrasonic scanning technology, three-dimensional data extraction and processing, and data fusion and three-dimensional model construction. In this embodiment, the specific implementation process of step S3 is described in detail as follows:

[0052] Step 3.1, ultrasonic probe precise positioning mechanism. After completing the uneven area length measurement in step S2, the control system immediately starts the ultrasonic probe precise positioning program to prepare for the subsequent three-dimensional scanning. This mechanism is intended to ensure that the ultrasonic probe can be accurately and stably positioned above the uneven area determined in step S2 and start scanning. The specific implementation process is as follows: First, the linear guide system receives the control instruction and drives the ultrasonic probe to move in a direction perpendicular to the surface of the steel plate to the preset scanning starting height H. The linear guide system is driven by a high-precision linear motor, combined with the closed-loop feedback of the photoelectric encoder to ensure that the positioning accuracy of the probe in the vertical direction reaches ±0.05mm. The setting of the scanning height H needs to comprehensively consider the working characteristics of the probe, the thickness of the steel plate and the expected maximum unevenness height, and is initially set according to the relationship:

[0053] H = H0 + max(10mm, 0.5D)

[0054] Where H0 is the reference scanning height, which is preset to 20 mm according to the probe model and working distance, and D is the maximum deformation measured in step S2. For example, if the maximum deformation D is 4 mm, the initial scanning height H = 20 mm + max(10 mm, 0.5 × 4 mm) = 30 mm.

[0055] Subsequently, the horizontal motion platform drives the ultrasonic probe to move in a plane parallel to the surface of the steel plate. The horizontal motion platform also uses a servo motor drive and high-precision encoder feedback to ensure that the positioning accuracy in the horizontal plane reaches ±0.1mm. According to the following relationship, the system calculates the optimal starting scanning position P0 of the ultrasonic probe:

[0056] P0=S1-max(20mm, 0.05L)

[0057] Among them, S1 is the starting coordinate of the uneven area determined in step S2, and L is the total length of the uneven area. The max(20mm, 0.05L) term in the formula is intended to ensure that the scanning starting position P0 can cover the boundary of the uneven area in advance, leaving a certain safety margin to avoid scanning and missing the edge information of the uneven area. For example, if the starting coordinate S1 of the uneven area is 100mm and the total length L is 300mm, the starting scanning position P0 = 100mm - max(20mm, 0.05 ×300mm) = 75mm. During the probe positioning process, the motion control system adopts a three-stage speed curve, including a fast approach stage, a precise positioning stage, and a working height adjustment stage to ensure the stability and accuracy of positioning. In order to further eliminate the influence of mechanical vibration on the measurement accuracy, the system sets a stabilization delay of 250ms after the probe reaches the target position to ensure that the ultrasonic scanning starts after the mechanical system is completely stable. Through the above-mentioned precise positioning mechanism of the ultrasonic probe, the ultrasonic probe can be accurately and stably positioned to the scanning starting position of the uneven area, providing a basis for subsequent high-precision stereo data acquisition.

[0058] Step 3.2, multi-spectrum phased array ultrasonic scanning technology. In order to achieve high-precision, all-round three-dimensional scanning of uneven areas, this embodiment adopts multi-spectrum phased array ultrasonic scanning technology. Compared with traditional single-frequency ultrasonic probes, phased array technology can achieve beam focusing and deflection by controlling the phase and delay of the array unit, thereby obtaining richer surface and internal information. Multi-spectrum technology can utilize the characteristics of ultrasonic waves of different frequencies to take into account the penetration depth and resolution of the scan. In this embodiment, the specific configuration of the phased array ultrasonic probe is as follows: the probe array includes 128 piezoelectric transducers arranged in a rectangular array with an array size of 60mm × 30mm. A single scan can cover an area larger than the typical seam width of the steel plate. The operating frequency range of each piezoelectric transducer unit is 3-10MHz, and the phase and power of the transmitted and received ultrasonic signals can be independently controlled. Ultrasonic scanning adopts a three-band alternating working mode: 1. Low frequency band (3-4MHz): Ultrasonic waves in this frequency band have strong penetration ability and are mainly used to obtain deep information of steel plates, for example, to detect whether there are defects or stratification inside the steel plate; 2. Medium frequency band (5-7MHz): Ultrasonic waves in this frequency band achieve a good balance between resolution and penetration, and are mainly used to collect the main contour data of uneven areas; 3. High frequency band (8-10MHz): Ultrasonic waves in this frequency band have high resolution and are mainly used to capture detailed features on the surface of steel plates, such as fine scratches or tiny bumps.

[0059] In the scanning mode, the system adopts an adaptive density strategy to dynamically adjust the scanning density D according to the following relationship: s :

[0060] D s =min[3mm, max(0.5mm, )

[0061] Where G is the deformation gradient of the local area, which reflects the speed of change of the surface unevenness. In the normal area (i.e., the flat area) determined in step S2, standard density scanning is adopted, and the scanning point spacing is set to 3mm to improve the scanning efficiency. In the area with large deformation, the system automatically increases the scanning density and reduces the scanning point spacing to 0.5mm to capture the morphological characteristics of the uneven area more finely. The max(0.5mm, ) ensures that the scanning density can be adaptively adjusted according to the deformation gradient G. When the deformation gradient G increases, the scanning density D sIt increases accordingly, but the minimum scanning spacing is limited to 0.5mm and the maximum scanning spacing is limited to 3mm. In addition, in order to obtain more comprehensive surface information, the system performs multi-angle focused beam scanning on each scanning point, including: 1. Center vertical beam (0°): used to accurately measure the height information of the scanning point, and the vertically incident ultrasonic wave can directly reflect the distance from the surface to the probe; 2. ±15° deflection beam: used to capture the reflected signal of the inclined surface and enhance the detection capability of the slope area; 3. ±30° deflection beam: used to detect areas with sharp surface changes, such as steep edges or deep pits. The deflection beam with a larger angle can better receive the reflected signal from these areas. To ensure continuous coverage of the steel plate joint area, a 30% overlap rate is maintained between adjacent scanning lines. The data in the overlapping area can not only improve the redundancy of the data, but also can be used for subsequent data splicing and calibration to improve the accuracy and reliability of 3D model construction. Through the above-mentioned multi-spectrum phased array ultrasonic scanning technology, the system can comprehensively and accurately collect three-dimensional morphological data of uneven areas, taking into account the depth, resolution and efficiency of the scan, and providing a high-quality data foundation for subsequent three-dimensional model construction and dynamic cutting parameter adjustment.

[0062] Step 3.3, 3D data extraction and processing. The original ultrasonic echo signal collected in step 3.2 contains rich morphological information of the uneven area, but it requires advanced signal processing technology to extract useful 3D data from it. In this embodiment, the system extracts four types of key data from the ultrasonic echo signal: concave and convex shape, height difference, precise boundary and surface curvature.

[0063] Step 3.3.1, concave and convex shape extraction. The system obtains the height distribution of the steel plate surface by analyzing the propagation time difference of the ultrasonic wave. First, the scanning data of the undeformed area (i.e., the flat area) is used as the reference plane to establish a three-dimensional coordinate system. For each scanning point, the system accurately extracts the time t1 of the first ultrasonic wave echo. Since the propagation speed v of the ultrasonic wave in the medium is known, the height Z of the scanning point relative to the reference plane can be calculated by the time-distance conversion formula: Z = Z0 ± v·(t1-t0) / 2, where Z0 is the reference plane height and t0 is the reference plane echo time. The positive and negative signs depend on the coordinate system definition and the probe measurement direction.

[0064] To improve the quality of height data, the system applies adaptive filtering algorithms to the raw height data, including: 1. Wavelet transform: used to remove high-frequency noise, such as electronic noise or random interference. 2. Median filtering: used to remove isolated outliers, such as noise caused by signal mutations or misjudgments. 3. Gaussian smoothing: used to optimize surface continuity and make the surface model smoother and more natural. After filtering, a high-precision 3D point cloud model of the uneven area is finally generated, and the resolution of the point cloud model reaches 0.1mm × 0.1mm × 0.05mm, that is, the plane resolution is 0.1mm × 0.1mm and the depth resolution is 0.05mm.

[0065] Step 3.3.2: Height difference measurement. Based on the 3D point cloud data generated in step 3.3.1, the system further extracts key height difference parameters to quantify the degree of deformation of the uneven area. The system first identifies and extracts key feature points from the point cloud data, including: 1. The highest point : The highest point of the uneven area; 2. The lowest point : The lowest point of the uneven area; 3. Average plane height : The average height calculated from the reference area (flat area) represents the ideal steel plate surface height.

[0066] The system then calculates several key height difference parameters:

[0067] 1. Maximum absolute height difference : Reflects the maximum deformation amplitude of the uneven area, and the calculation formula is: =| - |;

[0068] 2. Positive height difference relative to the reference surface : reflects the degree of protrusion of the uneven area relative to the ideal plane. The calculation formula is: = max(0, - );

[0069] 3. Negative height difference relative to the reference surface : reflects the degree of depression of the uneven area relative to the ideal plane. The calculation formula is: = max(0, - );

[0070] 4. Root mean square height deviation : Reflects the overall deformation degree of the uneven area. The calculation formula is: = ,in is the height of each point in the point cloud, and n is the total number of points in the point cloud.

[0071] The absolute measurement accuracy of the height difference data is ±0.05mm, and the relative measurement accuracy is better than 1% of the measured value, meeting the needs of high-precision cutting. Automatically classify the severity of uneven areas, for example: Level I (mild): < 0.5mm; Grade II (medium): 0.5mm ≤ < 2mm; Grade III (severe): ≥ 2mm.

[0072] Step 3.3.3, accurate boundary extraction. In order to accurately determine the actual boundary of the uneven area, the system uses gradient analysis technology. The system proposed above calculates the spatial derivative (gradient) G of the height data, which reflects the severity of the height change. When the height gradient G meets the following conditions, it is determined to be a boundary point: |G|>max(0.1mm / mm, 0.02× / mm), where h is the thickness of the steel plate. In the formula, max(0.1mm / mm, 0.02× / mm) item defines the dynamic boundary threshold. The size of the threshold is related to the thickness h of the steel plate. The greater the thickness, the higher the threshold is, so as to adapt to the surface characteristics of steel plates of different thicknesses.

[0073] Boundary extraction uses a closed contour tracking algorithm to ensure that the extracted boundary is continuous and closed. The algorithm process includes: first, multi-directional gradient analysis: calculate gradients from multiple directions to ensure that boundary changes in all directions are captured; second, sub-pixel boundary positioning: improve boundary positioning accuracy to sub-pixel level; third, B-spline curve fitting: use B-spline curve to fit boundary points to optimize the continuity and smoothness of the boundary. Finally, the boundary data is stored in vector form, including: 1. Boundary coordinate sequence ( , ): a sequence of point coordinates that constitute the boundary contour; 2. Boundary closed polygon: a closed polygon area formed by a sequence of boundary coordinates; 3. Feature point coordinates (inflection points, extreme points): coordinates of key feature points on the boundary contour.

[0074] Step 3.3.4, surface curvature calculation. To more comprehensively describe the surface morphology of the uneven area, the system calculates the surface curvature distribution based on the 3D point cloud model. The curvature reflects the degree of curvature of the surface and can more finely describe the concave and convex changes of the surface. For each grid point in the point cloud model, the system constructs a local quadratic surface for fitting, and then calculates the principal curvatures κ1 and κ2, as well as the average curvature . The curvature calculation uses multi-scale analysis, covering a feature size range of 1-10mm to accommodate surface features of different scales. To improve the quality of curvature data, the system applies adaptive smoothing technology to the curvature data to reduce noise while maintaining edge features. In areas with large curvature changes, more details are retained, while in areas with small curvature changes, smoothness is increased. Finally, the system generates a curvature distribution heat map to intuitively display the curvature information of the surface, for example: high curvature areas represent areas where the surface changes dramatically; zero curvature areas represent planes or transition areas; positive / negative curvature areas represent convex and concave areas, respectively.

[0075] Step 3.4, data fusion and three-dimensional model construction. In order to construct a complete three-dimensional model of the uneven area, the system uses multi-source data fusion technology to intelligently fuse the four types of key data extracted in step 3.3. The fused data includes: 1. Height data: providing absolute position information of the uneven area; 2. Boundary data: defining the scope of the uneven area; 3. Curvature data: enhancing the local morphological characteristics of the uneven area; 4. The lateral projection data provided in step S2: as auxiliary verification data, improving the reliability of the model.

[0076] Based on the fused data, the system constructs a 3D surface mesh model. The mesh density of the mesh model is adaptively matched to the surface complexity. In the complex concave-convex area, the mesh is encrypted and the mesh size is set to 0.2-0.5mm to finely describe the surface details; in the flat area, the mesh is sparse and the mesh size is set to 1-2mm to save data storage space and computing resources.

[0077] The constructed three-dimensional model contains complete semantic tags, and key feature points include the highest point, the lowest point, the saddle point, etc. The deformation severity zoning includes severity level areas divided according to the height difference parameter. The gradient change drastic area includes the boundary area. The symmetric and asymmetric areas include surface symmetry features analyzed according to the curvature distribution. The final generated three-dimensional model data set contains: 1. Geometric mesh data: including point coordinates, patch topological relationships, etc.; 2. Feature attribute data: including height, curvature, degree of deformation, etc.; 3. Metadata: including scanning parameters, data processing history, data quality assessment information, etc. Through the above data fusion and three-dimensional model construction process, the conversion of uneven areas from raw ultrasonic echo signals to structured three-dimensional models is realized, providing a complete and accurate spatial reference for the cutting path planning and dynamic focus adjustment in the subsequent step S4.

[0078] Step S4, precise cutting execution and dynamic focus adjustment. Step S4 is the core execution link of the present invention, and its purpose is to adjust the key parameters of the laser cutting head, including height, focal length and laser power, in real time and accurately during the laser cutting process according to the three-dimensional morphological data of the uneven area collected in step S3, so as to achieve high-quality and high-precision cutting of uneven areas of spliced ​​building steel plates. The core of this step lies in efficient and secure data transmission, intelligent cutting path planning, multi-dimensional collaborative dynamic focus adjustment, precise cutting execution and real-time monitoring. In this embodiment, the specific implementation process of step S4 is described in detail as follows:

[0079] Step 4.1, efficient and safe data transmission. After completing the three-dimensional data acquisition of the uneven area in step S3, the system immediately starts an efficient and safe data transmission mechanism, and reliably and quickly transmits the data set containing the complete three-dimensional information of the uneven area from the data acquisition unit to the laser cutting control system, providing a data basis for subsequent cutting path planning and dynamic parameter adjustment. The specific implementation process is as follows:

[0080] First, the system adopts a hierarchical structured data transmission scheme to organize the three-dimensional morphological data into a multi-level structure to meet the needs of control systems at different levels and improve data processing efficiency. The data hierarchy includes: 1. Basic layer data: Contains key macro parameters of the uneven area, such as the position coordinates, total length, and maximum height difference of the uneven area on the steel plate. These parameters provide the cutting control system with an overview of the uneven area for preliminary path planning and parameter setting. 2. Grid layer data: Contains height grid data formed by height sampling of the uneven area surface at regular intervals (for example, 0.5mm × 0.5mm or 1mm × 1mm). This layer of data provides detailed height information on the surface of the uneven area and is the core data source for dynamic focus adjustment. 3. Feature layer data: Contains the precise three-dimensional coordinates of key feature points on the surface of the uneven area, such as the highest point, the lowest point, and the point with the maximum curvature change rate. These feature point data are used to assist in cutting path optimization and fine-tuning cutting parameter adjustment in special areas. 4. Metadata layer: Contains auxiliary information such as parameter information of the data collection process, data accuracy assessment report, data generation timestamp, etc. Metadata is used for data traceability, quality control and system diagnosis.

[0081] Secondly, to ensure the high reliability and real-time performance of data transmission, the system adopts a dual-channel redundant data transmission mechanism, including: 1. Main channel: using Gigabit Ethernet direct connection, taking advantage of the high bandwidth and low latency characteristics of industrial Ethernet, to achieve high-speed data transmission, with a theoretical transmission rate of up to 980Mbps. The main channel undertakes the main data transmission tasks of the grid layer and feature layer. 2. Backup channel: using a dedicated high-speed serial interface (for example, RS-422 or fiber channel) as a backup for data transmission. When the main channel fails, the backup channel can automatically take over the data transmission task to ensure the reliable operation of the system. The backup channel has a transmission rate of 100Mbps and mainly transmits key control data such as the basic layer and metadata layer.

[0082] During the data transmission process, the system uses the CRC32 cyclic redundancy check algorithm to check the transmitted data packets to ensure the integrity and accuracy of the data during the transmission process. Any data packet that fails the check will be automatically retransmitted until the check succeeds, ensuring zero bit error rate of data transmission. In order to improve data processing efficiency, the cutting control system adopts a hierarchical data cache strategy. A multi-level data cache area is set up inside the system, and the total cache capacity is not less than 4GB. For the steel plate area to be cut, the system preloads the relevant data into the high-speed cache area in advance, and the data access delay is less than 1ms, ensuring the fast data access requirements of the real-time control system. For the data of the long-term cutting area, it is stored in the main cache area, and the data is dynamically scheduled according to the cutting progress to achieve efficient data management. After the data transmission is completed, the system executes the data check and preprocessing program. First, the data checksum of the data acquisition unit and the cutting control unit is compared to verify the integrity of the data transmission. Then, the data format conversion is performed to convert the collected data format into a coordinate system and data structure that can be directly recognized and used by the cutting system. Finally, a data ready signal is generated to notify the cutting control system that the data preparation is complete, and the cutting path planning and dynamic parameter adjustment can be started. Through the above-mentioned efficient and safe data transmission mechanism, the system can ensure that the 3D shape data of the uneven area is quickly, completely and reliably transmitted to the laser cutting control system, providing the data basis for subsequent precise cutting operations. The entire data transmission process (for a typical 1GB data volume) takes no more than 2 seconds, and the redundancy mechanism ensures high data transmission reliability.

[0083] Step 4.2, intelligent cutting path planning. After receiving the three-dimensional shape data of the uneven area transmitted in step S3, the cutting control system immediately starts the intelligent cutting path planning program. The core goal of this program is to generate an optimized laser cutting path based on the design drawings of the steel plate and the actual shape of the uneven area, and provide an accurate path reference for subsequent dynamic focus adjustment. The specific implementation process is as follows:

[0084] First, the system imports the design drawing data of the steel plate. The system is compatible with a variety of mainstream CAD design file formats, including: 1. Two-dimensional vector format: for example, DXF, IGES, etc., for importing two-dimensional cutting contour information; 2. Three-dimensional model format: for example, STEP, STP, etc., for importing three-dimensional part model information; 3. Rasterized image format: for example, BMP, JPEG, etc. For simple contour drawings, the system can perform vector processing and extract cutting path information. After importing the design drawing, the system performs position calibration between the design drawing and the actual steel plate. Since there may be slight position deviations in the conveying and positioning process of the steel plate, in order to ensure that the cutting path is accurately aligned with the actual position of the steel plate, the system adopts an automatic calibration method based on reference features. The reference points or edge features (for example, positioning holes, corner points, etc.) are marked on the steel plate in advance. The system obtains the actual position of these reference features through visual recognition or sensor detection, and then compares them with the reference positions in the design drawing to calculate the best registration transformation matrix (including translation, rotation and scaling transformation). The calibration accuracy is better than ±0.1mm to ensure the accuracy of the cutting position.

[0085] After completing the drawing calibration, the system maps the 3D data of the uneven area obtained in step S3 to the cutting path coordinate system. First, a unified 3D coordinate reference system is established, the surface of the steel plate is defined as the XY plane, and the direction perpendicular to the surface of the steel plate is defined as the Z axis. Then, the 3D point cloud data of the uneven area is transformed and aligned with the calibrated cutting path in a unified coordinate system. Next, the system calculates the intersection point set of the cutting path and the uneven area, and analyzes the situation where the cutting path passes through the uneven area.

[0086] According to the results of the cross-analysis, the system processes the cutting path in segments and adopts different cutting strategies for different areas: First, normal area (flat area): For areas where no unevenness is detected, the system adopts conventional laser cutting parameters, such as standard cutting speed, standard laser power, fixed focal length, etc.; Second, transition area: The area within a certain range (for example, 10-20mm) before and after the boundary of the uneven area is defined as the transition area. In the transition area, the system adopts a parameter smoothing transition strategy to gradually adjust the cutting parameters to ensure a smooth transition of the cutting quality and avoid sudden changes in cutting quality at the boundary of the uneven area. Third, uneven area: For the uneven area determined in step S2, the system applies a dynamic adjustment strategy to adjust the parameters of the laser cutting head in real time according to the three-dimensional morphological data collected in step S3.

[0087] For uneven areas, the system performs optimization calculations to provide detailed parameter curves for subsequent dynamic focus adjustments. Specifically, it includes: 1. Calculate the ideal focus position of each path point: According to the three-dimensional model of the uneven area, calculate the ideal focus Z-axis height of each point on the cutting path to ensure that the focus is always on the actual surface of the steel plate. 2. Generate the position-time curve of the electric lifting mechanism: According to the ideal focus position, generate a position-time curve to control the movement of the electric lifting mechanism to guide the dynamic movement of the cutting head in the Z-axis direction. 3. Generate the parameter-time curve for focal length adjustment: According to the ideal focus position and the change in the height of the cutting head, calculate the focal length adjustment parameters of the variable focal length optical system, generate the focal length parameter-time curve, and achieve fine focus position control. Fourth, generate the parameter-time curve for power adjustment: According to the change in focal length, calculate the compensation amount of laser power, generate the power parameter-time curve, and ensure constant energy density during the cutting process.

[0088] During the path planning process, the system also determines the cutting order and direction according to the following optimization rules to improve cutting efficiency and quality: 1. Prioritize cutting from the flat area into the uneven area: avoid cutting directly into the uneven area to reduce the instability at the beginning of the cutting. 2. Try to avoid turning in the high curvature area: in the high curvature area, the movement trajectory of the cutting head changes dramatically, which can easily affect the cutting quality. The turning operation should be minimized. 3. When there are multiple cutting paths, optimize the connection path and reduce the empty movement distance: for complex cutting contours, the system optimizes the connection order of the cutting path, reduces the empty movement distance of the cutting head in the non-cutting area, and improves the cutting efficiency.

[0089] Through the above intelligent cutting path planning mechanism, the system realizes the precise docking between the design drawings and the actual steel plate state, and generates an optimized cutting strategy for uneven areas, providing an accurate execution basis for subsequent dynamic focus adjustment. The path planning process is highly efficient. For a typical size of steel plate (for example, 2m × 1m), the path planning time does not exceed 5 seconds, meeting the real-time requirements of the production line.

[0090] Step 4.3, multi-dimensional collaborative dynamic focus adjustment. The multi-dimensional collaborative dynamic focus adjustment system is the core component of step S4 and is also the key innovation of the present invention. The goal of this system is to adjust multiple key parameters of the laser cutting head in real time and collaboratively during the laser cutting process, including Z-axis height, optical focal length and laser power, according to the three-dimensional morphological data provided by step S3 and the cutting path planned by step S4.2, to ensure that the laser focus always acts accurately on the cutting part of the uneven steel plate to achieve high-quality cutting. The dynamic focus adjustment system is mainly composed of three collaborative subsystems:

[0091] Electric lifting mechanism: used to control the height of the laser cutting head in the Z-axis direction to adjust the vertical position of the cutting head relative to the surface of the steel plate. In this embodiment, the electric lifting mechanism adopts high-precision linear drive technology, and the specific configuration is as follows: 1. Drive motor: AC servo motor with high response speed and high positioning accuracy; 2. Transmission mechanism: ball screw mechanism is adopted to achieve precise linear motion conversion; 3. Position sensor: high-resolution position encoder is adopted, with a position resolution of 0.005mm and a repeat positioning accuracy of ±0.01mm; 4. Dynamic performance: maximum speed response rate of 500mm / s, acceleration reaches 0.8G; 5. Travel range: vertical travel range is 50-100mm, which meets the adjustment needs of various uneven conditions.

[0092] Optical focal length adjustment mechanism: used to control the focusing parameters of the laser beam and achieve fine adjustment of the laser focal position. In this embodiment, the focus control system adopts a two-stage focusing mechanism: 1. Coarse adjustment: The overall height of the cutting head is adjusted by the electric lifting mechanism to achieve a large range of height changes. 2. Fine adjustment: The focus position is finely adjusted by a variable focal length optical system (for example, an adjustable focal length lens or a reflector) to achieve fine control of the focus position in a small range and at a high frequency. The adjustment range of the variable focal length optical system is ±2mm, and the response frequency is up to 50Hz.

[0093] Collaborative control: The electric lifting mechanism and the variable focal length optical system work together to achieve precise control of the focus position. The two-stage system works together according to the following relationship: ΔZ+ΔF=ΔH+C, where ΔZ is the height adjustment of the electric lifting mechanism, ΔF is the focal length adjustment of the variable focal length optical system, ΔH is the surface height change collected in step S3, and C is a constant correction term (usually ±0.05mm), which represents the correction bias of the system and is used to compensate for system errors.

[0094] Laser power adjustment system: It is used to dynamically compensate the laser power according to the change of focal length to ensure constant energy density during cutting. In this embodiment, the dynamic adjustment of laser power follows the following relationship:

[0095] P = P0 × [1 + k × ]

[0096] Among them, P is the adjusted laser power, P0 is the standard cutting power, k is the material correlation coefficient (for steel, the value range is usually 0.15-0.25, and it is accurately calibrated according to the steel grade and thickness), ΔF is the focal length change, and F0 is the reference focal length. This formula is derived based on the physical relationship between laser energy density and focal length, and can effectively compensate for energy density fluctuations caused by focal length changes. During the dynamic adjustment process, the system maintains the coordinated control of various parameters in real time to ensure the synchronous and coordinated changes of the cutting head position, focal length and laser power. The parameter adjustment frequency reaches 100-200Hz, which is much higher than the parameter adjustment frequency of conventional laser cutting systems (usually 10-20Hz). The time synchronization of each parameter adjustment is better than 0.5ms, and the spatial position synchronization accuracy is better than 0.02mm, ensuring that the laser focus is always on the ideal cutting plane. For high-speed cutting scenarios (cutting speed v exceeds the critical value , usually set to 30mm / s), the system adopts a feedforward control strategy to overcome the response delay of the control system. Based on the three-dimensional model of the uneven area, the system predicts the surface morphology in the next 0.1-0.2 seconds, calculates and caches the control instructions in advance, and eliminates the impact of control delay. Prediction time window The following relationship is satisfied:

[0097]

[0098] in, is the inherent delay time of the control system, is the system execution delay time, is the safety factor (usually in the range of 1.5-2.0). The feedforward control strategy is used when the cutting speed v exceeds the critical value. The critical value of this embodiment is Set to 30mm / s.

[0099] The above multi-dimensional collaborative dynamic focus adjustment mechanism can achieve precise matching of the laser focus and the uneven steel plate surface, ensuring that the laser energy always acts efficiently on the cutting part. The system can adapt to uneven areas with a height difference of ±5mm, and the dynamic adjustment accuracy is ±0.05mm, which is significantly better than the traditional fixed focus laser cutting system.

[0100] Step 4.4, precise cutting execution and real-time monitoring. Under the coordinated control of the dynamic focus adjustment system, the laser cutting head performs precise cutting operations according to the cutting path and dynamic parameters planned in step S4.2. At the same time, the system starts the real-time monitoring mechanism to comprehensively monitor the cutting process and fine-tune the parameters according to the monitoring results to ensure the stability and consistency of the cutting quality. The specific implementation process is as follows:

[0101] The laser cutting head performs the cutting operation according to the planned cutting path and dynamically adjusted parameters. In this embodiment, the laser cutting system is configured as follows: 1. Laser: A 6kW fiber laser is used to output a laser beam with a wavelength of 1064nm. 2. Cutting head: Equipped with a high-pressure oxygen-assisted cutting system, which uses high-pressure oxygen to blow away the molten metal to improve cutting efficiency and quality. 3. Cutting speed: Dynamically adjusted according to the thickness and unevenness of the steel plate, with an adjustment range of 10-80mm / s. During the cutting process, the real-time monitoring system monitors the key parameters and states of the cutting process in real time through multiple sensing mechanisms. The sensor system includes: 1. Photoelectric sensor: Installed near the cutting head to monitor the slit width and consistency in real time. By analyzing the changes in reflected light or transmitted light, it is determined whether the slit width meets the preset standard. 2. Infrared temperature sensor: Monitor the temperature distribution of the cutting area. By non-contact measurement of the temperature field in the cutting area, the heat input and heat-affected zone size of the cutting process are evaluated. 3. Acoustic sensor: Monitor the noise spectrum characteristics during the cutting process. By analyzing the frequency component and intensity of the cutting noise, the stability of the cutting process can be judged, for example, whether there is abnormal vibration or plasma instability. 4. Reflected light intensity sensor: monitors the intensity of the reflected light of the laser beam. The reflected light intensity is closely related to the laser focal position and the cutting surface state. By analyzing the change in the reflected light intensity, it can be determined whether the focal position has deviated. The monitoring system performs real-time evaluation of the cutting quality based on sensor data. The system defines a comprehensive cutting quality index Q, which is used to quantify the quality of the cutting. The calculation formula of the quality index Q is as follows:

[0102] Q=w1C+w2T+w3A+w4R

[0103] Where C is the slit width consistency score (calculated based on photoelectric sensor data), T is the temperature distribution rationality score (calculated based on infrared temperature sensor data), A is the acoustic feature regularity score (calculated based on acoustic sensor data), R is the reflected light intensity stability score (calculated based on reflected light intensity sensor data), w1 to w4 are weight coefficients, which are used to set the emphasis on cutting quality indicators according to different application scenarios to satisfy Σwᵢ = 1.

[0104] System preset cutting quality threshold When the real-time calculated cutting quality index Q is lower than the threshold When the system triggers the parameter fine-tuning program, it automatically adjusts the cutting parameters to improve the cutting quality. Dynamic threshold is used to make adaptive adjustments based on the unevenness of the local area. Dynamic threshold The calculation formula is as follows:

[0105] =0.85-0.05×( )

[0106] Among them, G is the unevenness measure of the local area (for example, the maximum height difference or curvature of the local area), and G0 is the reference value (usually set to 1mm, representing a medium degree of unevenness). This formula shows that the higher the unevenness, the lower the system's tolerance for cutting quality, allowing a lower quality threshold.

[0107] Parameter fine-tuning adopts an intelligent adaptive strategy. The system pre-establishes a sensitivity matrix of cutting quality and control parameters to describe the sensitivity of cutting quality indicators to changes in different control parameters (for example, laser power, cutting speed, focal position, etc.). Based on the real-time cutting quality evaluation results, the system calculates the optimal parameter adjustment direction and amplitude according to the sensitivity matrix to most effectively improve the cutting quality. The parameter fine-tuning amplitude is subject to certain restrictions to avoid system instability caused by excessive parameter adjustment. The parameter adjustment amplitude restrictions are as follows: 1. ≤0.15 (laser power adjustment is limited to ±15% of standard power); 2. ≤0.2 (cutting speed adjustment is limited to ±20% of standard speed); 3. ≤0.3mm (the height adjustment of the cutting head is limited to ±0.3mm).

[0108] Among them, ΔP is the laser power adjustment, P0 is the current laser power, ΔV is the cutting speed adjustment, V0 is the current cutting speed, and ΔZ is the cutting head height adjustment.

[0109] When the cutting is about to be completed, the system executes a smooth transition program. The system detects whether the cutting path is about to leave the uneven area. At 10-20mm away from the boundary of the uneven area, the system starts the parameter transition program, and smoothly returns the dynamically adjusted cutting parameters to the standard cutting parameters according to the S-shaped curve to ensure a smooth transition of the cutting parameters and avoid sudden changes in cutting quality at the boundary of the uneven area. After the parameter transition is completed, the system records the cutting execution log and sends a cutting completion signal to complete the entire cutting process. Through the above-mentioned precise cutting execution and real-time monitoring mechanism, high-quality cutting of uneven areas is achieved, and parameters can be fine-tuned according to real-time monitoring results to ensure the consistency and stability of cutting quality. Compared with the traditional open-loop control laser cutting system, this system has improved cutting accuracy and reduced scrap rate.

Claims

1. A laser cutting method for splicing building steel plates, characterized in that: The following steps are involved: Step S1, using a plurality of laser beams arranged along the length direction of the joint of the spliced ​​steel plates and projected at an inclined angle to illuminate the joint area, and judging whether there is unevenness at the joint by analyzing the misalignment and deformation of the light projection, if the light is misaligned and deformed, executing step S2; Step S2, switching the laser to a single horizontal light mode and moving it along the length direction of the seam, recording the starting and ending positions of the light deformation, and determining the length of the uneven area; Step S3, using an ultrasonic probe to scan the uneven area determined in step S2 to collect three-dimensional morphological data of the uneven area; Step S4, according to the three-dimensional morphological data collected in step S3, the height, focal length and power of the laser cutting head are adjusted in real time during the cutting process to achieve accurate cutting of uneven areas; Among them, the detection result of step S1 triggers the execution of step S2 and provides an initial scanning area, the uneven area range determined by step S2 provides a precise scanning area for step S3, and the three-dimensional morphological data obtained by step S3 provides a basis for parameter adjustment for step S4, so as to achieve precise control of the entire process from detection to cutting.

2. The laser cutting method for spliced ​​building steel plates according to claim 1, characterized in that: In step S1, the incident angle of the laser beam is Satisfy the formula: ， in, is the estimated uneven height of the steel plate, d is the joint width of the two steel plates; the spacing L of the laser array that generates the multiple laser beams satisfies the relationship: L≤ ， Where W is the minimum unevenness length that needs to be detected; The misalignment deformation of the light projection is determined by a dynamic threshold. When the offset Δd of the light relative to the ideal position satisfies the condition: Δd>max(0.15mm,0.02×h), it is determined to be an uneven area, where h is the thickness of the steel plate.

3. The laser cutting method for spliced ​​building steel plates according to claim 1, characterized in that: In step S2, when the laser is converted from the wide-area multi-light mode to the single transverse light mode, the multi-path light sources are converged into a single transverse light by a micro-rotating grating, and the same incident angle as step S1 is maintained; the moving speed v of the laser along the length direction of the seam satisfies the relationship: ， Where D is the length of the steel plate joint to be measured, and T is the maximum measurement time allowed by the system; The deformation determination adopts a boundary confirmation algorithm and records the conditions that meet: N consecutive frames (N≥3) of images are used as the boundary points of the uneven area, where: is the deformation measurement parameter, is the reference value of the deformation measurement parameter of the ideal plane, and h is the thickness of the steel plate.

4. The laser cutting method for spliced ​​building steel plates according to claim 1, characterized in that: In step S3, the starting scanning position P0 of the ultrasonic probe satisfies the relationship: P0=S1-max(20mm, 0.05L), Among them, S1 is the starting coordinate of the uneven boundary, and L is the total length of the uneven area; the ultrasonic probe is a phased array ultrasonic probe, which includes a rectangular array of 64-128 piezoelectric transducer units, and adopts a three-band alternating working mode of low frequency band (3-4MHz), medium frequency band (5-7MHz) and high frequency band (8-10MHz), and performs multi-angle focused scanning of the central vertical beam and ±15° and ±30° deflection beams for each scanning point.

5. The laser cutting method for spliced ​​building steel plates according to claim 4, characterized in that: In step S3, the scanning density D s Adopting adaptive strategy, satisfying the relationship: D s =min[3mm, max(0.5mm, ), where G is the deformation gradient of the local area; the uneven area boundary determination adopts gradient analysis technology, when the height gradient G meets the condition: |G|>max(0.1mm / mm, 0.02× / mm), it is determined as a boundary point, where h is the thickness of the steel plate; the system constructs a three-dimensional grid model through multi-source data fusion technology, the grid density is adaptively matched with the surface complexity, the grid in the complex concave and convex area is encrypted (0.2-0.5mm), and the grid in the flat area is sparse (1-2mm).

6. The laser cutting method for spliced ​​building steel plates according to claim 1, characterized in that: In step S4, the focus control system adopts a two-stage focusing mechanism, including a coarse adjustment of the cutting head height by an electric lifting mechanism and a fine adjustment of the focus position by a variable focal length optical system. The two-stage system works together to satisfy the relationship: ΔZ+ΔF=ΔH+C, Among them, ΔZ is the height adjustment amount, ΔF is the focal length adjustment amount, ΔH is the surface height change amount, and C is a constant correction term; the dynamic adjustment of the laser power of the laser cutting head satisfies the relationship: P=P0×[1+k× ], Among them, P is the adjusted power, P0 is the standard power, k is the material correlation coefficient, ΔF is the focal length change, and F0 is the reference focal length.

7. The laser cutting method for spliced ​​building steel plates according to claim 6, characterized in that: In step S4, a feedforward control strategy is adopted in the high-speed cutting scenario to predict the surface morphology in the next 0.1-0.2 seconds based on the three-dimensional model of the uneven area, calculate and cache the control instructions in advance, and predict the time window. Satisfies the relationship: ， in, To control the delay, For system execution delay, is the safety factor; the feedforward control is when the cutting speed v exceeds the critical value Automatically activated when 8. The laser cutting method for spliced ​​building steel plates according to claim 6, characterized in that: In step S4, the cutting process is monitored in real time by using a photoelectric sensor, an infrared temperature sensor, an acoustic sensor, and a reflected light intensity sensor, and a comprehensive cutting quality index is defined: Q=w1C+w2T+w3A+w4R, Among them, C is the consistency score of the slit width, T is the rationality score of the temperature distribution, A is the regularity score of the acoustic characteristics, R is the stability score of the reflected light intensity, and w1 to w4 are weight coefficients; when Q is lower than the threshold Trigger parameter fine-tuning, =0.85-0.05×( ), where G is the local area unevenness measure, G0 is the reference value; the parameter fine-tuning amplitude satisfies the condition: ≤0.15, ≤0.2, ≤0.3mm, where ΔP is the power adjustment, ΔV is the speed adjustment, and ΔZ is the height adjustment.

9. The laser cutting method for spliced ​​building steel plates according to claim 1, characterized in that: In the step S1, an industrial-grade high-speed CCD camera system with a resolution of not less than 2048×1536 pixels is used, and a narrow-band filter with a central wavelength of 650nm±15nm is equipped to form a three-dimensional monitoring system; in the step S2, the frame rate of the camera system is increased to 120-240 frames / second, and the real-time image processing unit adopts a parallel computing architecture, and the processing delay is controlled within 5 milliseconds; between the steps S3 and S4, a dual-channel redundant mechanism is used to transmit the three-dimensional morphological data, the main channel adopts a Gigabit Ethernet direct connection, and the backup channel adopts a dedicated serial interface to ensure transmission reliability.

10. A cutting production line structure, used to implement the laser cutting method of spliced ​​building steel plates according to claim 1, characterized in that: include: A preliminary detection module, including a laser irradiation unit and a preliminary deformation analysis unit, is used to irradiate the joint area and analyze the dislocation deformation, determine whether there is unevenness, and output an unevenness trigger signal; The length measurement module includes a laser scanning unit and a length determination unit, which scans the joint area after receiving the unevenness trigger signal, records the start and end positions of the deformation, and outputs the length data of the uneven area; The three-dimensional data acquisition module includes an ultrasonic scanning unit and a morphological data construction unit, receives the range data of the uneven area, scans and constructs the three-dimensional morphological data of the uneven area; Dynamic cutting module, including laser cutting unit and cutting parameter adjustment unit, adjusts the height, focal length and power of the laser cutting head according to the three-dimensional morphological data to achieve precise cutting of uneven areas; Among them, the detection results of the preliminary detection module trigger the length measurement module, which determines the range of the uneven area and provides an accurate scanning area for the three-dimensional data acquisition module. The collected data provides a basis for real-time cutting parameter adjustment for the dynamic cutting module, thereby achieving precise cutting control.

Images