A laser cutting method for splicing building steel plates and the structure of a cutting production line

Abstract

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 splicing building steel plates, including a "preliminary rapid detection" step S1, an "uneven area length measurement" step S2, a "three-dimensional data acquisition" step S3, and a "cutting link" step S4. The execution of step S2 is triggered by the detection result of step S1 and an initial scanning area is provided. The range of the uneven area determined by step S2 provides an accurate scanning area for step S3. The three-dimensional shape 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. The purpose of the present invention is to solve the problem of potential safety hazards that in the prior art, when cutting and splicing building steel plates, especially when there is unevenness at the joint, the cutting accuracy is significantly reduced, which may lead to a decrease in the strength and stability of the building structure.

Description

Technical Field

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

[0002] In modern architecture, steel structures are widely used due to their excellent performance, and the spliced steel plate structure is 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, existing laser cutting technology has limitations when processing spliced building steel plates and is difficult to meet the production requirements of high quality and high efficiency. Traditional laser cutting technology mainly targets single flat steel plates. However, in actual production lines, especially when continuously processing spliced steel plates, the following technical bottlenecks are faced:

[0003] 1. Processing problems at the splicing joint: Traditional processes ignore the unevenness and thickness changes at the splicing joint caused by processes such as welding. This makes the cutting parameters set for flat steel plates no longer applicable, resulting in a decline in the cutting quality and efficiency at the splicing joint. Existing technologies are insufficient in continuously and highly quality cutting at the splicing joint, affecting production efficiency and material utilization rate. 2. General unevenness of steel plates: Building steel plates are prone to deformation during production, transportation, and installation, resulting in unevenness such as warping and bending. This unevenness makes it difficult to accurately control the relative position between the steel plate surface and the laser cutting head. Traditional fixed-focus laser cutting heads are difficult to adapt to the undulations 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 easily deviated from the actual surface of the steel plate, exceeding the effective cutting range. This causes the laser energy to be unable to effectively act on the cutting part, reducing the cutting efficiency and quality, and even possibly resulting in cutting failure. Especially at key splicing parts, a reduction in accuracy may affect the safety of the building structure. Existing technologies are difficult to accurately match uneven steel plates, restricting the application of laser cutting.

[0004] Generally speaking, existing laser cutting technology faces challenges such as processing blanks at the splicing joint of spliced building steel plates, poor adaptability to uneven steel plates, insufficient accuracy, and lack of real-time monitoring. Solving the problems of dynamic matching and accuracy improvement of the laser cutting head at the splicing joint of uneven steel plates is a key technical problem that needs to be solved urgently, especially in the context of the increasing requirements for safety and quality in the construction industry. Summary of the Invention

[0005] In order to solve the above technical problems of the existing technology to a certain extent as much as possible, the present invention provides a laser cutting method and a cutting production line structure for splicing building steel plates, aiming to solve the problem that when cutting spliced building steel plates in the existing technology, especially when there is unevenness at the joint, the cutting accuracy is significantly reduced, which may lead to potential safety hazards such as a decline in the strength and stability of the building structure.

[0006] The present invention discloses a laser cutting method for splicing building steel plates. This method first executes step S1, namely the "preliminary rapid detection" step. In this step, the system utilizes a plurality of laser beams linearly arranged along the length direction of the joint of the splicing steel plate and projects them onto the joint area of the steel plate at a preset inclination angle. The reason for using a plurality of laser beams and inclined projection is to quickly and sensitively capture possible uneven deformations at the joint. The inclined projection of the light will form regular projections on a flat surface. Once there are height differences or angular deviations at the joint, the light projection will be misaligned and deformed. The system uses image analysis technology to continuously monitor the deformation of these light projections in real time. If the analysis result shows that the light projection has a misalignment 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 requires 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 steps can be skipped (although the handling method for the flat situation is not clearly specified in this embodiment, in practical applications, direct standard cutting or re-verification can be considered).

[0007] In step S2, the "uneven area length measurement" step, the system switches the working mode of the laser to a single horizontal light beam mode. Different from the rapid screening with multiple light beams in step S1, step S2 aims to precisely measure the length range of the uneven area. The single horizontal light beam 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 beam and records the starting position where the light beam starts to have obvious deformation 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 result of step S2 provides the length range information of the target area for the precise scanning in the subsequent step S3.

[0008] Step S3, the "three-dimensional data acquisition" step, is a key preparatory step for achieving precise cutting. After determining the length range of the uneven area in the previous step, step S3 uses an ultrasonic probe to scan this area to obtain the three-dimensional morphological data of the uneven area. The ultrasonic probe can non-contactedly and precisely measure the height information of the object surface. By densely scanning the uneven area, the system can construct a three-dimensional point cloud model or depth map of this area. These three-dimensional morphological data precisely describe the key geometric information such as the unevenness, height changes, etc. of the uneven area, providing a data basis for the dynamic cutting parameter adjustment in the subsequent step S4. The three-dimensional morphological data obtained in step S3 is the parameter basis for precisely cutting the uneven area.

[0009] Finally, step S4, the "cutting 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 various 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 topography of the uneven area, aiming 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 undulation, so as to achieve precise cutting of the uneven area. The dynamic parameter adjustment in step S4 is the actuator for realizing precise cutting.

[0010] According to a laser cutting method for splicing building steel plates of the present invention, in the step S1, the incident angle of the laser beam satisfies the relationship:

[0011]

[0012] where is the estimated uneven height of the steel plate, and d is the joint width of two steel plates. It can be understood that the design of the incident angle of the laser beam θ = 45°+ arctan(h1 / 2d) optimizes the geometric relationship between the laser beam and the steel plate surface. When the light is incident on the uneven surface at this angle, there will be an offset with a displacement of Δd = 2×Δh×sin(θ). This angle design achieves the best balance between detection sensitivity and signal quality: the larger the angle, the more obvious the light displacement caused by the unevenness of the same height, improving the detection sensitivity; however, 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 according to the estimated uneven height h1 of the steel plate and the joint width d, significantly improving the detection accuracy and sensitivity, and reducing the missed detection rate.

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

[0014] 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(0.15 mm, 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 misalignment deformation of the light is determined by the dynamic threshold Δd > max(0.15 mm, 0.02×h), which solves the problem of inconsistent detection standards for steel plates of different thicknesses. The basic threshold of 0.15 mm ensures the detection ability of the system for minute unevenness. At the same time, the dynamic threshold of 0.02×h takes into account the factor of the steel plate thickness, because the manufacturing tolerance of thicker steel plates is usually larger, and it is necessary to appropriately increase the detection threshold. This threshold relationship can reduce the false positive rate of the system and at the same time ensure that the detection rate of unevenness above 0.2 mm is improved, meeting the quality control requirements for the 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 splicing building steel plates of the present invention, in the step S2, when the laser changes from the wide-area multi-light mode to the single transverse light mode, a micro-rotating grating is used to converge multiple light sources into a single transverse light, and the incident angle is kept the same as that in step S1; the moving speed v of the laser along the seam length direction satisfies the relationship: , where D is the length of the seam of the steel plate to be measured, and T is the maximum measurement time allowed by the system; the deformation determination adopts a boundary confirmation algorithm to record the consecutive N frames (N≥3) of images that satisfy the condition: as the boundary points of the uneven area, where, is the deformation metric parameter, is the reference value of the deformation metric parameter of the ideal plane, and h is the thickness of the steel plate. It can be understood that the above solution involves the working mode conversion, moving speed control and deformation boundary determination method of the laser in step S2. The technical effects produced by these features are as follows: the technical feature that the laser changes from the wide-area multi-light mode to the single transverse light mode realizes a seamless switch from "wide-area rapid screening" to "precision directional measurement". By using a micro-rotating grating to converge multiple light sources into a single transverse light and keeping the same incident angle as that in step S1, this design ensures the consistency and continuity of the detection results, with a short conversion time and reduced mode switching time. This efficient optical path reconstruction method avoids the inconvenience of traditional equipment replacement or long-time adjustment, improving the system response speed and detection efficiency. In addition, the moving speed v of the laser along the seam length direction is v = min[50 mm / s, The design achieves 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 allowable measurement time T. When the boundary of the uneven area shows a gradual change characteristic, the scanning speed can be reduced to improve the boundary positioning accuracy, while in the area with a clear boundary, the speed can be increased to improve the detection efficiency. This dynamic speed control strategy reduces the overall detection time, which is crucial for continuous operation on the production line. Finally, the deformation determination adopts a boundary confirmation algorithm, recording N consecutive frames (N ≥ 3) of images that satisfy the condition of |η - η0| > max(0.15, 0.05× ) as the boundary points of the uneven area. This design significantly improves the accuracy and anti-interference ability of boundary recognition. The introduction of the square root function (0.05× ) instead of the non-linear function better adapts to the non-linear relationship between the steel plate thickness and the manufacturing accuracy. The determination requirement of N consecutive frames effectively filters out environmental interference and instantaneous misjudgments, enabling the system to maintain a high boundary recognition accuracy rate on steel plates of various thicknesses in the range of 5 - 50 mm, while significantly reducing the misjudgment rate. This technical effect provides an accurate target area for subsequent three-dimensional data acquisition and improves the measurement accuracy of the length of the uneven area.

[0016] A laser cutting method for splicing building steel plates according to the present invention. In step S3, the starting scanning position P0 of the ultrasonic probe satisfies the relational expression: P0 = S1 - max(20mm, 0.05L), where 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 composed of 64 - 128 piezoelectric transducer units, and adopts an alternating working mode in three frequency bands: low frequency band (3 - 4MHz), medium frequency band (5 - 7MHz), and high frequency band (8 - 10MHz), and performs multi-angle focused scanning of the center vertical beam and ±15°, ±30° deflected beams on each scanning point. It can be understood that the determination of the starting scanning position of the ultrasonic probe in step S3, the configuration and working mode of the phased array ultrasonic probe. The technical effects produced by 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 exceeding the uneven area by 20mm or 5% of its total length), effectively avoiding subsequent cutting problems that may be caused by missing boundary data. This intelligent buffer design has an adaptive characteristic, 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 various situations, and improving the reliability and integrity of the measurement. The phased array ultrasonic probe adopts a rectangular array containing 64 - 128 piezoelectric transducer units, realizing the regional scanning ability with high spatial resolution. The array coverage area (60mm × 30mm) is larger than the typical seam width of the steel plate, ensuring that the seam area can be completely covered by a single scan. The scanning accuracy reaches ±0.05mm in the depth direction and ±0.2mm in the plane direction, far superior to traditional single-frequency ultrasonic detection technology. The alternating working mode in three frequency bands (low frequency 3 - 4MHz, medium frequency 5 - 7MHz, high frequency 8 - 10MHz) makes full use of the complementary characteristics of ultrasonic waves with different frequencies: the low frequency provides deep penetration ability, the high frequency provides fine surface details, and the medium frequency balances the two. This design enables the system to obtain comprehensive three-dimensional information in a single scan, and the detection ability 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 fine surface changes, it is increased to 8 - 10MHz to capture fine details. And the multi-angle focused beam scanning (center vertical beam and ±15°, ±30° deflected beams) solves the "blind area" problem of traditional single vertical incident ultrasonic detection when facing an inclined surface. For a surface with an inclined angle of α, when the detection beam angle θ satisfies |θ - α| < 15°, the best detection effect can be obtained.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), enabling the system to automatically adjust the boundary determination criteria for steel plates of different thicknesses, maintain a high boundary recognition accuracy rate on steel plates of various thicknesses within the range of 5 - 50 mm, and at the same time reduce the misjudgment rate. This precise boundary recognition technology provides an accurate basis for regional division for subsequent cutting parameter adjustment. In addition, the multi-source data fusion technology constructs a three-dimensional grid model, and the grid density is adaptively matched with the surface complexity, achieving efficient data representation. The grid is encrypted in complex concave-convex change areas (0.2 - 0.5 mm) to ensure high-precision expression of key feature areas; the grid is sparsified in flat areas (1 - 2 mm), significantly reducing the overall complexity of the model. This adaptive grid density design enables the system to significantly reduce the data volume while maintaining the model accuracy, on average, reducing more storage space, and at the same time improving the data processing and transmission efficiency. The generated optimized three-dimensional model not only retains the precise expression of key geometric features but also greatly reduces the data processing burden, providing an ideal data basis for subsequent cutting path planning and parameter optimization, directly affecting the final cutting quality and efficiency.

[0018] According to a laser cutting method for splicing building steel plates of the present invention, in the step S4, the focus control system adopts a two-stage focusing mechanism, including a coarse adjustment for adjusting the height of the cutting head through an electric lifting mechanism and a fine adjustment for finely adjusting the focus position through a variable focal length optical system. The two-stage system works together to satisfy the relationship: ΔZ + ΔF = ΔH + C, where 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 × , where P is the adjusted power, P0 is the standard power, k is the material-related coefficient, ΔF is the change in focal length, and F0 is the reference focal length. It can be understood that the focus control system adopts a two-stage focusing mechanism, including a coarse adjustment for adjusting the height of the cutting head through an electric lifting mechanism and a fine adjustment for finely adjusting the focus position through a variable focal length optical system. The two-stage system works together to satisfy the relationship: ΔZ + ΔF = ΔH + C. This design achieves precise focus control over the entire range: the coarse adjustment mechanism provides a large range of adjustment capabilities (50 - 100 mm) to adapt to uneven height changes of various scales; the fine adjustment mechanism provides fine adjustment capabilities with high frequency (50 Hz) and high precision (±2 mm) to ensure precise focus positioning. When the two work together, the system can quickly respond to the surface height change ΔH and achieve the best dynamic response characteristics by reasonably allocating the mechanical height adjustment ΔZ and the optical focal length adjustment ΔF. The constant correction term C (usually ±0.05 mm) provides a system error compensation function, offsetting the inherent deviations of the mechanical system and the optical system and improving the accuracy and stability of the overall system. In addition, this two-stage focusing mechanism enables the system to adapt to uneven areas with a height difference of up to ±5 mm, with a dynamic adjustment accuracy of ±0.05 mm, solving the problem of the focus position deviation during cutting in uneven areas by traditional fixed-focus systems, resulting in a significant decrease in the cutting energy density. The cutting accuracy can also be significantly improved, and the cutting quality consistency is significantly enhanced. 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 physical characteristics of laser optics. By accurately calculating the influence of the focal length change on the focal spot size (the focal spot size A ∝ (1 + |ΔF / F0|)²) and adjusting the laser power accordingly, it ensures that the energy density at the focus remains constant during the focal length adjustment, achieving a consistent cutting effect. Introducing the material-related coefficient k (usually 0.15 - 0.25 for steel) enables the system to automatically adjust the power compensation coefficient according to different material characteristics, improving the system's adaptability to different types of steel. This power dynamic adjustment mechanism solves the problem of unstable cutting quality in traditional fixed-power systems when the focal length changes, significantly improving the cutting depth consistency, the surface quality stability, and at the same time improving the energy efficiency. The system can achieve a cutting accuracy of ±0.1 mm in uneven areas. Even in areas with a height difference of up to ±5 mm, the change in the slit width is still controlled within the range of ±0.05 mm, providing a key technical guarantee for the manufacture of high-quality building steel structures.

[0019] According to a laser cutting method for splicing building steel plates of the present invention, in step S4, a feedforward control strategy is adopted in a high-speed cutting scenario. Based on the three-dimensional model of the uneven area, the surface morphology in the next 0.1 - 0.2 seconds is predicted, and the control instructions are calculated and cached in advance. The prediction time window satisfies the relationship: , where is the control delay. is the system execution delay, is the safety factor; the feedforward control is automatically activated when the cutting speed v exceeds the critical value , and the critical value is set to 30 mm / s. It can be understood that in the high-speed cutting scenario, the feedforward control strategy is adopted to predict the surface morphology in the next 0.1 - 0.2 seconds based on the three-dimensional model of the uneven area, and calculate and cache the control instructions in advance. This design solves the "tracking error" problem of the traditional feedback control system during high-speed cutting, where the actual adjustment action lags behind the surface change due to control delay. During high-speed cutting, the tracking error Δe, the cutting speed v, the control delay τ, and the surface gradient G satisfy the relationship: Δe = v × τ × G. The feedforward control effectively compensates for this delay by predicting the future surface morphology and executing adjustment actions in advance, making the control action precisely synchronized with the actual demand, and controlling the dynamic adjustment error within ±0.1 mm, improving the dynamic accuracy by about 70% compared with the traditional feedback control. The prediction time window satisfies the relational expression: , where τ is the control delay, is the system execution delay, is the safety factor. 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 adapting to different system configurations and operating states. The safety factor (usually 1.5 - 2.0) provides an additional time margin, enhancing the system's adaptability to delay fluctuations. Even when the system load changes or temporary interference factors appear, effective feedforward control can still be maintained. The design that the feedforward control is automatically activated when the cutting speed v exceeds the critical value (set to 30 mm / s) realizes the optimal allocation of control resources. During low-speed cutting, ordinary feedback control can meet the accuracy requirements, and only the computationally intensive feedforward control is activated during high-speed cutting, avoiding unnecessary complex calculations. The clear setting of the speed critical value enables the system to automatically select the most suitable control strategy according to the cutting speed, realizing a smooth transition from feedback control to feedforward control and avoiding the instability that may be brought about by the switching of control strategies.

[0020] According to a laser cutting method for splicing building steel plates of the present invention, in step S4, the cutting process is monitored in real time through 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, where C is the score of the seam width consistency, T is the score of the temperature distribution rationality, A is the score of the acoustic feature regularity, R is the score of the reflected light intensity stability, and w1 to w4 are weight coefficients; when Q is lower than the threshold Trigger parameter fine-tuning when = 0.85 - 0.05×( ), where G is the unevenness measure of the local area and G0 is the reference value; the parameter fine-tuning range satisfies the condition: ≤ 0.15, ≤ 0.2, ≤ 0.3 mm, where ΔP is the power adjustment amount, ΔV is the speed adjustment amount, and ΔZ is the height adjustment amount. It can be understood that the cutting process is monitored in real time through a photoelectric sensor, an infrared temperature sensor, an acoustic sensor, and a reflected light intensity sensor, and a comprehensive cutting quality index Q = w1C + w2T + w3A + w4R is defined. This multi-sensor fusion monitoring system realizes a comprehensive evaluation of the cutting quality: the photoelectric sensor monitors the slit width and consistency, the infrared temperature sensor monitors the temperature distribution in the cutting area, the acoustic sensor monitors the spectral characteristics of the cutting noise, and the reflected light intensity sensor monitors the focal position deviation. Multiple sensors work together to form a comprehensive monitoring of the cutting quality, which can capture complex quality problems that are difficult to identify by a single sensor, significantly improving the accuracy and comprehensiveness of the quality evaluation. Different sensors have different sensitivities to various cutting anomalies. Multi-sensor fusion enables the system to detect potential anomalies early before the problems seriously affect the cutting quality. Compared with single-sensor monitoring, the advance amount of anomaly detection is significantly improved on average. The cutting quality threshold = 0.85 - 0.05×( )'s design realizes an adaptive quality standard. The basic threshold of 0.85 represents the minimum cutting quality standard required by the system in normal flat areas. The dynamic adjustment term -0.05×(G / G0) introduces the unevenness degree G as a variable, setting reasonable quality expectations for different degrees of uneven areas. Flatter areas maintain a higher standard (close to 0.85), and highly uneven areas appropriately relax the standard while still ensuring basic quality requirements, making the quality control both strict and realistic. This dynamic threshold mechanism avoids the problem of excessive adjustment in high-difficulty cutting areas, reduces the number of unnecessary parameter adjustments, improves the system stability by about 35%, and at the same time ensures timely intervention when necessary to maintain quality balance. In addition, the amplitude of parameter fine-tuning meets the conditions: |ΔP / P0|≤0.15, |ΔV / V0|≤0.2, |ΔZ|≤0.3mm. By setting the maximum amplitude limit for each parameter adjustment, it avoids system oscillations or out-of-control situations that may be caused by excessive adjustment, ensuring that even in extreme cases, the adjustment actions remain within the safe control range. Limiting the single adjustment amplitude prompts the system to improve the cutting quality in a small-step and progressive manner rather than a large-scale mutation. This way optimizes the parameters step by step while maintaining the cutting continuity, ultimately achieving a better comprehensive effect. Compared with large-scale adjustments, the generated cutting marks have a smoother transition. At the same time, by limiting the adjustment amplitude of each parameter and considering the comprehensive influence of multiple parameters, the system can find the optimal combination of power, speed, and height. This collaborative optimization improves the effectiveness of parameter adjustment. In summary, these technical features together achieve intelligent monitoring and precise adjustment of the cutting process, enabling the system to maintain high-quality cutting in complex uneven areas. Compared with traditional open-loop control, the cutting accuracy of this system is significantly improved, and the scrap rate is also significantly reduced. The combination of closed-loop real-time monitoring and adaptive parameter adjustment enables the system to handle various complex cutting scenarios, including changes in steel plate thickness, material inhomogeneity, etc., ensuring the stability and consistency of cutting quality, and providing technical guarantees for the safety and reliability of building steel structures.

[0021] According to a laser cutting method for splicing 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 adopted, equipped with a narrow-band filter with a central wavelength of 650nm±15nm to form a three-dimensional monitoring system; in step S2, the frame rate of the camera system is increased to 120 - 240 frames per second, and the real-time image processing unit adopts a parallel computing architecture with the processing delay controlled within 5 milliseconds; between step S3 and step S4, a dual-channel redundancy mechanism is adopted to transmit three-dimensional form data. The main channel is directly connected by gigabit Ethernet, and the standby channel adopts a dedicated serial interface to ensure transmission reliability.

[0022] The present invention also discloses a cutting production line structure for implementing the laser cutting method for splicing 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 is 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, records the starting and ending positions of the deformation, and outputs the length data of the uneven area after receiving the unevenness trigger signal; a three-dimensional data acquisition module, including an ultrasonic scanning unit and a morphological data construction unit, which scans and constructs the three-dimensional morphological data of the uneven area after receiving the range data of the uneven area; a dynamic cutting module, including a laser cutting unit and a cutting parameter adjustment unit, which adjusts the height, focal length, and power of the laser cutting head according to the three-dimensional morphological data to achieve precise cutting of the uneven area. Among them, the detection result of the preliminary detection module triggers the length measurement module, determines the range of the uneven area and provides an accurate scanning area for the three-dimensional data acquisition module, and the acquired data provides a basis for real-time cutting parameter adjustment of the dynamic cutting module, thereby realizing precise cutting control.

[0023] The technical effects that can be achieved by the laser cutting method for splicing building steel plates of the present invention are as follows: 1. Improve cutting accuracy: Through the precise detection and three-dimensional data acquisition system of four steps (preliminary rapid detection, length measurement of the uneven area, three-dimensional data acquisition, and cutting link), accurately obtain the morphological information of the uneven area at the joint. According to this information, dynamically adjust the parameters of the laser cutting head to ensure that the laser beam always acts precisely on the steel plate surface, significantly improving the cutting accuracy of the uneven area. 2. Enhance the structural safety of the steel plate: High-precision cutting and high-quality cut seams, especially at the key splicing parts of building steel plates, 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. Improve production efficiency: Adopt a combination of preliminary rapid detection and precise measurement, and only carry out fine processing on the uneven area, while the flat area can still adopt an efficient conventional cutting mode. The dynamic adjustment process is automated, reducing manual intervention and improving the overall production efficiency. 4. Improve material utilization rate: Precise cutting reduces waste products and rework, especially when cutting complex contours or special-shaped spliced parts, making the most of the steel plate material, reducing material waste, improving the material utilization rate, and reducing production costs. These technical effects together solve the problem of low cutting accuracy in the prior art when cutting and splicing building steel plates, especially when there is unevenness at the joint, thereby avoiding potential safety hazards that may lead to a decrease in the strength and stability of the building structure. Brief Description of the Drawings

[0024] Figure 1 is the flowchart of the steps of the present invention. Detailed Embodiments

[0025] 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 and are not intended to limit the protection scope of the present invention.

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

[0027] Step S1, preliminary rapid detection. Step S1 aims 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 project a multi-ray laser light array at an inclined angle onto the joint area of the spliced steel plate, and capture the deformation of the light projection through a high-speed industrial camera system, so as to quickly judge 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 the steel plate. First, use a high-precision conveying system to convey the spliced steel plate to be cut to the detection station. The conveying system adopts a multi-group anti-slip coating synchronous conveying roller driven by a servo motor to ensure that the steel plate is conveyed forward along the production line at a stable speed of 2 - 5 meters per minute. To achieve precise positioning, an infrared sensor array is set on both sides of the production line to monitor the position of the steel plate 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 meter per minute. Further, a precision electronic distance measurer is set 2 meters upstream of the detection area to continuously monitor the moving distance of the steel plate. When the distance measurer determines that the leading edge of the steel plate reaches the preset coordinate of the leading edge of the detection position, the control system issues a stop signal, and the conveying roller stops rotating, completing the preliminary longitudinal positioning of the steel plate. After the steel plate stops, start the four-corner hydraulic positioner to apply a 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, effectively avoiding the impact on the overall efficiency of the production line.

[0028] Through the above precise conveying and positioning mechanism, the spliced steel plate can be quickly and accurately positioned to the detection area, providing an accurate initial position for subsequent laser detection and ensuring the accuracy and consistency of the detection results.

[0029] Step 1.2, Laser system configuration and light projection. After completing the steel plate positioning, start the laser system for light projection. In this embodiment, a multi-channel laser array is used, and 20 semiconductor laser emitters are installed at equal intervals along the length direction of the steel plate seam. Each laser emitter outputs visible red light with a wavelength of 650nm ± 10nm, and the diameter of a single light spot is 1.2mm. The entire laser array is installed on a precision bracket that can rotate 360°, and the incident angle θ of the laser light relative to the steel plate surface is adjusted through 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 seam width d between two steel plates to ensure the detection sensitivity. In this embodiment, the maximum estimated uneven height h of the steel plate is 5mm, the seam width d is 1mm, and according to the formula:

[0030]

[0031] The incident angle θ is calculated to be approximately 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 too large an incident angle of the light.

[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 size of uneven areas. The spacing L of the laser array that generates the multiple laser beams satisfies the relationship:

[0033] L ≤

[0034] Among them, the minimum uneven length W to be detected is determined to be 50mm according to the building standard, and the incident angle θ is 60°. The maximum spacing L of the laser array is calculated to be approximately 8.3mm. To ensure the detection reliability, in this embodiment, the spacing L of the laser array is set to 8mm, ensuring that each laser beam can span the seam between two adjacent steel plates, and the light density is sufficient to detect the smallest size of uneven areas. The power of the laser light source is controlled at 4mW, which can not only ensure that the light projection is clearly visible but also pose no safety hazard to the steel plate surface or the operator. Through the above laser system configuration and light projection, a uniform laser beam array with a certain incident angle can be formed in the splicing steel plate seam area, providing the necessary optical signals for subsequent image acquisition and unevenness recognition.

[0035] Step 1.3, Camera Monitoring System. To capture the changes of the laser light projection on the steel plate joint in real time, a high-speed industrial camera system is adopted in this embodiment. Two industrial-grade high-speed CCD cameras are installed 2.5 meters above the steel plate, and the two cameras are arranged at a 15° angle to form a three-dimensional monitoring system, with the field of view covering the entire joint area. The camera resolution is 2048×1536 pixels, and the frame rate is 100 frames per second to ensure that minute light deformations can be captured. The focal length of the camera lens is 25mm, the aperture is F2.8, and it is equipped with a narrowband filter with a central wavelength of 650nm±15nm, effectively enhancing the recognition rate of the laser light and filtering out ambient stray light interference. The camera system adopts a hard trigger mode to achieve precise synchronization with the laser, with a trigger delay of 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 through Gigabit Ethernet, with a transmission delay of less than 8 milliseconds, meeting the requirements of real-time processing. The above high-speed industrial camera system can achieve high-precision three-dimensional imaging of the steel plate joint area, providing a reliable data basis for subsequent image analysis and unevenness recognition.

[0036] Step 1.4, Light Misalignment Recognition and Judgment. After receiving the images transmitted by the camera system, the image processing unit uses a specially developed image processing algorithm for light misalignment recognition and judgment. First, the image processing unit extracts the projection positions of each laser light on the steel plate surface and establishes a baseline reference model, which represents the projection positions of the laser lights on the steel plate surface in an ideal flat state. Subsequently, the system analyzes the shape and position of the projection of each laser light in real time and detects its offset Δd relative to the baseline reference model.

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

[0038] Δd>max(0.15mm,0.02×h)

[0039] Where h is the nominal thickness of the steel plate. In this embodiment, the thickness of the steel plate is 10mm, and the calculated judgment threshold is Δd > max(0.15mm, 0.2mm) = 0.2mm. That is, when the light offset exceeds 0.2mm, the system determines that there is unevenness in this area. The time response of the system to determine unevenness does not exceed 80 milliseconds, ensuring the real-time nature of the detection. For the 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 in the subsequent step S2.

[0040] Step S2: Measuring the length of the uneven area. Step S2 is initiated after it is preliminarily detected in Step S1 that there is unevenness at the joint. 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 three-dimensional data acquisition in the subsequent Step S3, and provide a length basis for the final adjustment of the dynamic cutting parameters. 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 the accurate identification of the boundaries 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 it is determined in Step S1 that there is unevenness at the joint, the control system immediately starts the working mode conversion program, switching the laser system from the wide-area multi-ray rapid detection mode to the single transverse ray precise measurement mode. This conversion process aims to focus the laser light source for more refined scanning measurement. Specifically, the control unit sends a mode switching signal to the laser array, and the built-in multiplexer in the system completes the state switching within 3 milliseconds. The optical path switching device in the laser assembly is then activated, using a micro-rotating grating to converge the originally multi-path divergent laser beam into a single transverse ray. The diameter of this single transverse ray is precisely adjusted to 1.0 mm to improve the measurement resolution. Importantly, during the mode switching process, the single transverse ray maintains the same inclined incident angle θ (60° in this embodiment) as the laser ray in Step S1. To ensure the accuracy of the angle, the system starts a precision calibration program, using a preset reference calibration block to verify the light projection angle. 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 is restored within the allowable range. After the mode switching is completed, to improve the signal-to-noise ratio of the single ray, the laser emission power is automatically increased to 9 mW, but still controlled within the power range safe for the steel plate surface and the operator. The entire mode conversion process is completed within 180 milliseconds, achieving a rapid and seamless switch of the detection mode, ensuring the continuity and efficiency of the detection process. Through the above working mode conversion mechanism, the laser system can quickly switch from the wide-area detection to the fine measurement mode, providing the necessary light source conditions for the subsequent measurement of the length of the uneven area.

[0042] Step 2.2 High-precision motion control system. To achieve the precise scanning of a single transverse laser beam along the seam length direction, this embodiment adopts a high-precision motion control system. The core component of this system is a telescopic device, which integrates a main body telescopic mechanism and a precision displacement platform inside to jointly achieve the accurate positioning and stable movement of the laser. The main body telescopic mechanism uses a synchronous toothed belt drive driven by a servo motor, which is responsible for the rough positioning of the laser in the seam length direction, and its positioning accuracy reaches ±0.8 mm. The precision displacement platform is driven by a linear motor and has sub-micron positioning ability, with a repeat positioning accuracy better than ±0.008 mm, and is responsible for the fine displacement control during the scanning process. Among them, the scanning speed v of the laser along the seam length direction adopts a dynamic adjustment strategy, satisfying the relationship:

[0043]

[0044] where D is the length of the seam of the steel plate to be measured (estimated from 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 seam length D is 1000 mm, the calculated scanning speed v = min[50 mm / s, (1000 mm / (10 * 20 s))] = 5 mm / s.

[0045] The motion control system adopts a three-segment speed curve, including a smooth acceleration stage, a constant-speed scanning stage, and a smooth deceleration stage. The acceleration and deceleration are both controlled at 0.4g to reduce inertial shock and vibration. To further eliminate the influence of mechanical vibration on the measurement accuracy, the guide rail system is equipped with an active vibration damping device, which uses a piezoelectric ceramic actuator to offset the vibration in real time. When the amplitude exceeds 0.02 mm, the vibration damping device is immediately activated. The high-resolution position encoder real-time feedbacks the position information of the laser, with a resolution of 0.003 mm, ensuring that the system accurately grasps the spatial coordinates of each measurement point. Through the above high-precision motion control system, the stable and accurate scanning of a single transverse laser beam along the seam length direction can be achieved, providing a reliable motion platform for subsequent image acquisition and uneven area boundary recognition.

[0046] Step 2.3 Real-time Image Acquisition and Processing. During the laser scanning process, the camera monitoring system simultaneously 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 blurring caused by the movement of the laser. The camera field of view is adjusted to the narrowband focusing 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 + 30 mm, where S is the distance between the farthest adjacent uneven points detected in step S1. In addition, the image acquisition system adopts the polling cache technology and sets up a three-level cache queue to ensure that no image frame is lost during the high-speed acquisition process. The real-time image processing unit executes the following processing flow: S1. Narrowband Filtering: Apply the narrowband filtering algorithm to further filter out ambient stray light and accurately extract the contour of the laser beam; S2. Curvature and Offset Calculation: Calculate the actual projection curvature K and position offset D of the laser beam at each acquisition point. The curvature reflects the degree of light beam deformation, and the offset reflects the height of the unevenness; S3. Generation of Deformation Measurement Parameters: Compare with the ideal light beam model (straight line) to generate the deformation measurement parameter η = f(K, D). This parameter comprehensively reflects the degree of light beam deformation, and the larger the value, the more serious the deformation; S4. Data Association Storage: Associate and store the deformation measurement parameter η with the timestamp and spatial position information to form an ordered measurement data stream. To meet the real-time requirement, the image processing unit adopts a parallel computing architecture, processes the image in blocks, and controls the processing delay within 4 milliseconds. The above real-time image acquisition and processing mechanism can capture the subtle changes in the laser beam projection at high speed and accurately, and convert the light beam deformation information into quantifiable deformation measurement parameters, providing key data for subsequent identification of the boundary of the uneven area and calculation of the length.

[0047] Step 2.4 Boundary Identification and Length Calculation. In this embodiment, an accurate boundary identification algorithm is adopted 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 the boundary determination condition:

[0048]

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

[0050] The system records the starting coordinate S1 and the ending coordinate 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 the safety margin, which is set to 4mm in this embodiment to ensure complete coverage of the uneven area. For uneven areas with complex shapes, the system can identify multiple consecutive sub-areas, and the length Li of each sub-area needs to meet the requirement of the minimum effective uneven area length Lmin, which is set to 8mm in this embodiment. 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 ending point S2); 2. The total length L of the uneven area; 3. The severity grading (according to the maximum value of the deformation metric parameter η, for example, divided into three levels: slight, medium, and severe); 4. The spatial distribution curve of the deformation parameters within the area. This report will be used as the basis for the scanning range of the three-dimensional data acquisition in step S3 and provide the length parameter for the subsequent cutting path planning.

[0051] Step S3: Three-dimensional data acquisition. Step S3 is started based on the accurate measurement of the length of the uneven area in step S2, aiming to acquire the three-dimensional morphological data of the uneven area with high precision, providing 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, the multi-spectrum phased array ultrasonic scanning technology, the extraction and processing of three-dimensional data, and the 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: Precise positioning mechanism of the ultrasonic probe. After completing the measurement of the length of the uneven area in step S2, the control system immediately starts the precise positioning program of the ultrasonic probe to prepare for the subsequent three-dimensional scanning. This mechanism aims to ensure that the ultrasonic probe can accurately and stably be 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 along the direction perpendicular to the steel plate surface to the preset scanning starting height H. This linear guide system is driven by a high-precision linear motor and combines 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 uneven height, and is initially set according to the relationship:

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

[0054] Wherein, H0 is the reference scanning height, 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, then 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 within a plane parallel to the surface of the steel plate. The horizontal motion platform also uses a servo motor drive and a high-precision encoder feedback to ensure that the positioning accuracy in the horizontal plane reaches ±0.1 mm. According to the following relationship, the system calculates the optimal starting scanning position P0 of the ultrasonic probe:

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

[0057] Wherein, S1 is the starting coordinate of the uneven area determined in step S2, and L is the total length of the uneven area. The term max(20 mm, 0.05L) in the formula is designed 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 missing the edge information of the uneven area during scanning. For example, if the starting coordinate S1 of the uneven area is 100 mm and the total length L is 300 mm, then the starting scanning position P0 = 100 mm - max(20 mm, 0.05 × 300 mm) = 75 mm. During the probe positioning process, the motion control system adopts a three-stage speed curve, including a rapid approach stage, a precise positioning stage, and a working height adjustment stage, to ensure the smoothness and accuracy of positioning. To further eliminate the influence of mechanical vibration on the measurement accuracy, the system sets a stable delay of 250 ms after the probe reaches the target position to ensure that the mechanical system is completely stable before starting the ultrasonic scan. Through the above ultrasonic probe precise positioning mechanism, the ultrasonic probe can be accurately and stably positioned at the scanning starting position of the uneven area, providing a basis for subsequent high-precision three-dimensional data acquisition.

[0058] Step 3.2, Multi-spectrum phased array ultrasonic scanning technology. To achieve high-precision and omni-directional 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 control the phase and delay of array elements to achieve beam focusing and deflection, thereby obtaining richer surface and internal information. The multi-spectrum technology can utilize the characteristics of ultrasonic waves with different frequencies to balance the penetration depth and resolution of scanning. In this embodiment, the specific configuration of the phased array ultrasonic probe is as follows: The probe array consists of 128 piezoelectric transducer units, 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 transmitting and receiving ultrasonic signals can be independently controlled. The ultrasonic scanning adopts a three-band alternating working mode: 1. Low-frequency band (3-4MHz): Ultrasonic waves in this band have strong penetration ability and are mainly used to obtain deep information of the steel plate. For example, detecting whether there are defects or delaminations inside the steel plate; 2. Medium-frequency band (5-7MHz): Ultrasonic waves in this 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 band have high resolution and are mainly used to capture the detailed features of the steel plate surface, such as fine scratches or tiny bumps and depressions.

[0059] In the scanning mode, the system adopts an adaptive density strategy and dynamically adjusts the scanning density D according to the following relational expression s :

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

[0061] where G is the deformation gradient of the local area, reflecting the change rate of the surface unevenness. In the normal area (i.e., 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 areas with a large degree of deformation, the system automatically increases the scanning density, and the scanning point spacing is reduced to 0.5mm to more finely capture the morphological characteristics of the uneven area. The max(0.5mm, ) term in the formula 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 an adaptive filtering algorithm to the original 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 points 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 three-dimensional 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 planar resolution is 0.1mm × 0.1mm and the depth resolution is 0.05mm.

[0065] Step 3.3.2, height difference measurement. Based on the three-dimensional point cloud data generated in Step 3.3.1, the system further extracts key height difference parameters to quantify the deformation degree 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 position point of the uneven area; 2. The lowest point : the lowest position point of the uneven area; 3. The average plane height : the average height calculated from the reference area (flat area), representing the ideal steel plate surface height.

[0066] Then, the system calculates multiple key height difference parameters:

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

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

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

[0070] 4. The root mean square height deviation : reflects the overall deformation degree of the uneven area, and the calculation formula is: = , where 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 above height difference data reaches ±0.05 mm, and the relative measurement accuracy is better than 1% of the measured value, meeting the requirements of high-precision cutting. The system also automatically divides the severity level of the uneven area according to the maximum absolute height difference into severity levels, for example: Level I (slight): < 0.5 mm; Level II (medium): 0.5 mm ≤ < 2 mm; Level III (severe): ≥ 2 mm.

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

[0073] The 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 the gradient from multiple directions to ensure that all boundary changes in all directions are captured; Second, sub-pixel level boundary localization: improve the boundary localization accuracy to reach the sub-pixel level; Third, B-spline curve fitting: use B-spline curves to fit the 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 ( , ): The sequence of point coordinates that make up the boundary contour; 2. Boundary closed polygon: The closed polygon area formed by the boundary coordinate sequence; 3. Feature point coordinates (inflection points, extreme points): The 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 three-dimensional point cloud model. Curvature reflects the degree of surface bending and can more precisely describe the concave and convex changes on 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, and the mean curvature The curvature calculation adopts multi-scale analysis, covering a feature size range of 1 - 10 mm to adapt to surface features of different scales. To improve the quality of curvature data, the system applies adaptive smoothing technology to the curvature data, reducing noise while preserving edge features. In regions with large curvature changes, more details are retained, while in regions with small curvature changes, the smoothness is increased. Finally, the system generates a heat map of the curvature distribution to visually display the curvature information of the surface. For example, high-curvature regions represent areas with drastic surface undulations; zero-curvature regions represent flat or transitional regions; positive / negative curvature regions represent convex and concave regions respectively.

[0075] Step 3.4, Data Fusion and 3D Model Construction. To construct a complete 3D model of the uneven region, 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 the absolute position information of the uneven region; 2. Boundary data: defining the scope of action of the uneven region; 3. Curvature data: enhancing the local morphological features of the uneven region; 4. Lateral projection data provided in Step S2: serving as auxiliary verification data to improve 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. The mesh is densified in regions with complex undulations, and the mesh size is set to 0.2 - 0.5 mm to finely describe the surface details; the mesh is sparsified in flat regions, and the mesh size is set to 1 - 2 mm to save data storage space and computing resources.

[0077] The constructed 3D model contains complete semantic tags. Key feature points include the highest point, the lowest point, saddle points, etc. The severity partition of deformation includes severity level regions divided according to the height difference parameter. The region with drastic gradient changes includes the boundary region. The symmetric and asymmetric regions include the surface symmetry features analyzed based on the curvature distribution. The finally generated 3D model dataset includes: 1. Geometric mesh data: including point coordinates, patch topological relationships, etc.; 2. Feature attribute data: including height, curvature, deformation degree, etc.; 3. Metadata: including scanning parameters, data processing history, data quality assessment information, etc. Through the above data fusion and 3D model construction process, the conversion of the uneven region from the original ultrasonic echo signal to a structured 3D model is realized, providing a complete and accurate spatial reference for the cutting path planning and focal point dynamic adjustment in the subsequent Step S4.

[0078] Step S4, Precision Cutting Execution and Dynamic Focus Adjustment. Step S4 is the core execution link of the present invention. Its purpose is to adjust the key parameters of the laser cutting head, including height, focal length, and laser power, in real time and precisely during the laser cutting process according to the three-dimensional shape data of the uneven area collected in Step S3, so as to achieve high-quality and high-precision cutting of the uneven area of the spliced building steel plate. The core of this step lies in efficient and safe data transmission, intelligent cutting path planning, multi-dimensional collaborative dynamic focus adjustment, as well as precision 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 the three-dimensional data collection of the uneven area is completed in Step S3, the system immediately activates an efficient and safe data transmission mechanism to reliably and quickly transfer the data set containing the complete three-dimensional information of the uneven area from the data collection 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 shape data into a multi-level structure to meet the requirements of different levels of control systems and improve data processing efficiency. The data levels include: 1. Basic layer data: containing the key macroscopic 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 an overview of the uneven area for the cutting control system and are used for preliminary path planning and parameter setting. 2. Grid layer data: containing height grid data formed by height sampling of the surface of the uneven area at regular intervals (e.g., 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: containing the precise three-dimensional coordinates of the key feature points on the surface of the uneven area, such as the highest point, the lowest point, and the point with the largest curvature change rate. These feature point data are used to assist in optimizing the cutting path and adjusting the refined cutting parameters for special areas. 4. Metadata layer: containing auxiliary information such as the parameter information of the data collection process, the data accuracy assessment report, and the data generation timestamp. Metadata is used for data traceability, quality control, and system diagnosis.

[0081] Secondly, to ensure high reliability and real-time performance of data transmission, the system adopts a dual-channel redundant data transmission mechanism, including: 1. Main channel: It uses the direct connection method of Gigabit Ethernet. By leveraging the high bandwidth and low latency characteristics of industrial Ethernet, high-speed data transmission is achieved, and the theoretical transmission rate can reach 980 Mbps. The main channel undertakes the main data transmission tasks of the grid layer and feature layer. 2. Spare channel: It uses a dedicated high-speed serial interface (such as RS-422 or Fibre Channel) as a backup for data transmission. When the main channel fails, the spare channel can automatically take over the data transmission task to ensure the reliable operation of the system. The transmission rate of the spare channel is 100 Mbps, and it 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 verify the transmitted data packets to ensure the integrity and accuracy of the data during transmission. Any data packet with a verification failure will be automatically retransmitted until the verification is successful, ensuring a zero error rate for data transmission. To improve data processing efficiency, the cutting control system adopts a hierarchical data caching strategy. Multiple levels of data buffer areas are set inside the system, and the total buffer capacity is not less than 4 GB. For the steel plate area to be cut soon, the system preloads the relevant data into the high-speed buffer area in advance, with a data access latency of less than 1 ms to meet 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 buffer 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 verification and preprocessing program. First, compare the data verification sums of the data acquisition unit and the cutting control unit to verify the integrity of the data transmission. Then, perform data format conversion to convert the acquired data format into the coordinate system and data structure that the cutting system can directly recognize and use. Finally, generate a data ready signal to notify the cutting control system that the data preparation is completed and the cutting path planning and dynamic parameter adjustment can start. Through the above efficient and secure data transmission mechanism, the system can ensure that the three-dimensional shape data of the uneven area is quickly, completely, and reliably transmitted to the laser cutting control system, providing a data basis for subsequent precise cutting operations. The entire data transmission process (for a typical data volume of 1 GB) takes no more than 2 seconds, and the redundancy mechanism ensures high reliability of data transmission.

[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 drawing 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 formats: For example, DXF, IGES, etc., which are used to import two-dimensional cutting contour line information; 2. Three-dimensional model formats: For example, STEP, STP, etc., which are used to import three-dimensional part model information; 3. Raster image formats: For example, BMP, JPEG, etc. For simple contour drawings, the system can perform vectorization processing to 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 during the transportation and positioning of the steel plate, to ensure the precise alignment of the cutting path with the actual position of the steel plate, the system adopts an automatic calibration method based on reference features. Benchmark points or edge features (such as positioning holes, corner points, etc.) are pre-marked on the steel plate. The system obtains the actual positions 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 optimal registration transformation matrix (including translation, rotation, and scaling transformations). The calibration accuracy is better than ±0.1mm, ensuring the accuracy of the cutting position.

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

[0086] According to the cross-analysis results, the system processes the cutting path in segments and adopts different cutting strategies for different regions: First, the normal area (flat area): For the area 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, the 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 smooth 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, the uneven area: For the uneven area determined in step S2, the system applies a dynamic adjustment strategy and adjusts the parameters of the laser cutting head in real time according to the three-dimensional shape data collected in step S3.

[0087] For uneven areas, the system performs optimization calculations to provide a detailed parameter curve for subsequent dynamic focus adjustment. Specifically, it includes: First, calculating the ideal focus position of each path point: According to the 3D 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 located on the actual surface of the steel plate. Second, generating the position-time curve of the electric lifting mechanism: Based on the ideal focus position, generate the position-time curve for controlling the movement of the electric lifting mechanism to guide the dynamic movement of the cutting head in the Z-axis direction. Third, generating the parameter-time curve of 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 and generate the focal length parameter-time curve to achieve precise focus position control. Fourth, generating the parameter-time curve of power adjustment: According to the change in focal length, calculate the compensation amount of laser power and generate the power parameter-time curve to ensure a constant energy density during the cutting process.

[0088] During the path planning process, the system also determines the cutting sequence and direction according to the following optimization rules to improve cutting efficiency and quality: 1. Prioritize cutting into the uneven area from the flat area: Avoid directly cutting into the uneven area to reduce instability at the initial stage of cutting. 2. Try to avoid turning in areas with high curvature: In areas with high curvature, the movement trajectory of the cutting head changes drastically, which easily affects the cutting quality. Therefore, turning operations should be minimized. 3. When there are multiple segments of cutting paths, optimize the connection path to reduce the idle travel distance: For complex cutting profiles, the system optimizes the connection sequence of the cutting paths to reduce the idle travel distance of the cutting head in the non-cutting area and improve cutting efficiency.

[0089] Through the above intelligent cutting path planning mechanism, the system realizes the precise docking between the design drawing and the actual state of the steel plate, generates an optimized cutting strategy for the uneven area, and provides an accurate execution basis for subsequent dynamic focus adjustment. The path planning process has high computational efficiency. For steel plates of typical sizes (e.g., 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 also the key innovation point of the present invention. The goal of this system is to adjust multiple key parameters of the laser cutting head, including the Z-axis height, optical focal length, and laser power, in real-time and collaboratively during the laser cutting process according to the 3D shape data provided in step S3 and the cutting path planned in step S4.2, ensuring that the laser focus always precisely acts on the cutting part of the uneven steel plate to achieve high-quality cutting. The dynamic focus adjustment system mainly consists of three subsystems that work together:

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

[0092] Optical focal length adjustment mechanism: It is used to control the focusing parameters of the laser beam and realize the fine adjustment of the laser focus position. In this embodiment, the focus control system adopts a two-stage focusing mechanism: 1. Coarse adjustment: Adjust the overall height of the cutting head through the electric lifting mechanism to achieve a large range of height changes for adaptation. 2. Fine adjustment: Fine-tune the focus position through a variable focal length optical system (for example, an adjustable focal length lens or mirror) 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 ±2 mm, and the response frequency is as high as 50 Hz.

[0093] Coordinated control: The electric lifting mechanism and the variable focal length optical system work together to jointly 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 amount of the electric lifting mechanism, ΔF is the focal length adjustment amount of the variable focal length optical system, ΔH is the surface height change amount collected in step S3, and C is a constant correction term (usually ±0.05 mm), representing the correction bias of the system, which is used to compensate for system errors.

[0094] Laser power adjustment system: It is used to dynamically compensate the laser power according to the focal length change to ensure a constant energy density during the cutting process. In this embodiment, the dynamic adjustment of the 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-related coefficient (for steel, the general value range is 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 the energy density fluctuation caused by focal length change. During the dynamic adjustment process, the system maintains the coordinated control of each parameter in real time to ensure the synchronous and coordinated change 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 the conventional laser cutting system (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 located on the ideal cutting surface. For high-speed cutting scenarios (the 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. The system is based on the three-dimensional model of the uneven area, predicts the surface morphology in the next 0.1 - 0.2 seconds, calculates and caches the control instructions in advance, and eliminates the influence of control delay. The prediction time window satisfies the following relationship:

[0097]

[0098] Among them, is the inherent delay time of the control system, is the execution delay time of the system, is the safety factor (the general value range is 1.5 - 2.0). The feedforward control strategy is automatically activated when the cutting speed v exceeds the critical value , and the critical value of this embodiment is set to 30mm / s.

[0099] The above multi-dimensional collaborative dynamic focus adjustment mechanism can achieve the 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 up to ±5mm, and the dynamic adjustment accuracy reaches ±0.05mm, which is significantly better than the traditional fixed-focus laser cutting system.

[0100] Step 4.4, Precision 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 activates the real-time monitoring mechanism to comprehensively monitor the cutting process, and performs parameter fine-tuning 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 cutting operations 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 to blow away molten metal using high-pressure oxygen, improving 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 uses a multiple sensing mechanism to monitor the key parameters and status of the cutting process in real time. The sensor system includes: 1. Photoelectric sensor: Installed near the cutting head to monitor the cutting seam width and consistency in real time. By analyzing the changes in reflected light or transmitted light, it is judged whether the cutting seam width meets the preset standard. 2. Infrared temperature sensor: Monitors the temperature distribution in the cutting area. By non-contact measurement of the temperature field in the cutting area, the heat input and the size of the heat affected zone during the cutting process are evaluated. 3. Acoustic sensor: Monitors the noise spectrum characteristics during the cutting process. By analyzing the frequency components and intensity of the cutting noise, the stability of the cutting process is judged. For example, whether there are abnormal vibrations or plasma instabilities. 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 focus position and the cutting surface state. By analyzing the changes in the reflected light intensity, it can be judged whether the focus position deviates. The monitoring system performs real-time evaluation of the cutting quality based on the sensor data. The system defines a comprehensive cutting quality index Q 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 cutting seam width consistency score (calculated based on the photoelectric sensor data), T is the temperature distribution rationality score (calculated based on the infrared temperature sensor data), A is the acoustic feature regularity score (calculated based on the acoustic sensor data), R is the reflected light intensity stability score (calculated based on the reflected light intensity sensor data), and w1 to w4 are weight coefficients, which are set according to the emphasis on the cutting quality indicators in different application scenarios, satisfying Σwᵢ = 1.

[0104] The system presets a cutting quality threshold . When the real-time calculated cutting quality index Q is lower than the threshold , the system triggers a parameter fine-tuning program to automatically adjust the cutting parameters to improve the cutting quality. The cutting quality threshold adopts a dynamic threshold and is adaptively adjusted according to the unevenness of the local area. The calculation formula of the dynamic threshold is as follows:

[0105] = 0.85 - 0.05×( )

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

[0107] The parameter fine-tuning adopts an intelligent adaptive strategy. The system pre-establishes a sensitivity matrix of cutting quality and control parameters, which describes the sensitivity of cutting quality indicators to changes in different control parameters (e.g., laser power, cutting speed, focus 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 limiting conditions to avoid system instability caused by excessive parameter adjustment. The limiting conditions for parameter adjustment amplitude are as follows: 1. ≤ 0.15 (the adjustment amount of laser power is limited within ±15% of the standard power); 2. ≤ 0.2 (the adjustment amount of cutting speed is limited within ±20% of the standard speed); 3. ≤ 0.3 mm (the adjustment amount of the cutting head height is limited within ±0.3 mm).

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

[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 a distance of 10 - 20 mm from the boundary of the uneven area, the system starts the parameter transition program, smoothly returning the dynamically adjusted cutting parameters to the standard cutting parameters according to the S-shaped curve, ensuring a smooth transition of the cutting parameters and avoiding 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 precise cutting execution and real-time monitoring mechanism, high-quality cutting of the uneven area is achieved, and parameter fine-tuning can be performed according to the real-time monitoring results to ensure the consistency and stability of cutting quality. Compared with the traditional open-loop control laser cutting system, the cutting accuracy of this system is improved and the scrap rate is reduced.

Claims

1. A laser cutting method for splicing building steel plates, characterized in that, Including the following steps: Step S1: Irradiate the joint area with multiple laser beams arranged along the length direction of the splicing steel plate seam and projected at an inclined angle. Determine whether there is unevenness at the joint by analyzing the misalignment deformation of the light projection. If the light shows misalignment deformation, execute Step S2; Step S2: Switch the laser to a single transverse light mode and move it along the length direction of the seam. Record the starting and ending positions of the light deformation to determine the length of the uneven area; Step S3: Use an ultrasonic probe to scan the uneven area determined in Step S2 and collect the three-dimensional morphological data of the uneven area; Step S4: According to the three-dimensional morphological data collected in Step S3, adjust the height, focal length, and power of the laser cutting head in real time during the cutting process to achieve precise cutting of the uneven area; Among them, the execution of Step S2 is triggered by the detection result of Step S1 and an initial scanning area is provided. The range of the uneven area determined in Step S2 provides an accurate scanning area for Step S3. The three-dimensional morphological data obtained in Step S3 provides a basis for parameter adjustment in Step S4 to achieve precise control of the entire process from detection to cutting; In Step S4, the focus control system adopts a two-stage focusing mechanism, including a coarse adjustment for adjusting the height of the cutting head through an electric lifting mechanism and a fine adjustment for finely adjusting the focus position through a variable focal length optical system. The two-stage system works together to satisfy the relationship: ΔZ + ΔF = ΔH + C, where 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 of the laser cutting head is dynamically adjusted to satisfy the relationship: P = P0 × [1 + k × , where P is the adjusted power, P0 is the standard power, k is a material-related coefficient, ΔF is the focal length change amount, and F0 is the reference focal length.

2. The laser cutting method for splicing building steel plates according to claim 1, characterized in that In the step S1, the incident angle of the laser beam satisfies the following relationship: ， Among them, is the estimated uneven height of the steel plate, and 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 uneven length that needs to be detected; The misalignment deformation of the light projection is determined by a dynamic threshold. When the offset amount Δd of the light relative to the ideal position satisfies the condition: Δd > max(0.15 mm, 0.02×h), it is determined as an uneven area, where h is the thickness of the steel plate.

3. The laser cutting method for splicing building steel plates according to claim 1, wherein, In the step S2, when the laser is converted from a wide-area multi-ray mode to a single transverse ray mode, a micro-rotating grating is used to converge multiple light sources into a single transverse ray, and the incident angle is kept the same as that in step S1; the moving speed of the laser along the seam length direction satisfies the relation: ， where D is the length of the seam of the steel plate to be measured, and T is the maximum measurement time allowed by the system; The deformation determination adopts a boundary confirmation algorithm, recording consecutive N frames of images that meet the condition: as the boundary points of the uneven area, where N ≥ 3, 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 splicing building steel plates according to claim 1, characterized in that In the step S3, the starting scanning position P of the ultrasonic probe a satisfies the relational expression: P a = S1 - max(20 mm, 0.05L), where 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 composed of 64 - 128 piezoelectric transducer units and adopts an alternating working mode in three frequency bands: low frequency band, medium frequency band, and high frequency band. The frequency of the low frequency band is 3 - 4 MHz, the frequency of the medium frequency band is 5 - 7 MHz, and the frequency of the high frequency band is 8 - 10 MHz. Multi-angle focused scanning of the center vertical beam and ±15°, ±30° deflection beams is performed for each scanning point.

5. The laser cutting method for splicing building steel plates according to claim 4, wherein In the step S3, the scanning density D s Adopts an adaptive strategy to satisfy the relationship: D s = min[3mm, max(0.5mm, ), where G is the deformation gradient of the local area; the boundary determination of the uneven area adopts gradient analysis technology. When the deformation gradient G of the local area satisfies 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 mesh model through multi-source data fusion technology, and the mesh density is adaptively matched with the surface complexity. The mesh is encrypted in the area with complex concave and convex changes, and the mesh size after encryption is 0.2 - 0.5mm. The mesh is thinned in the flat area, and the mesh size after thinning is 1 - 2mm.

6. The laser cutting method for splicing building steel plates according to claim 1, characterized in that, In step S4, a feedforward control strategy is adopted in the high-speed cutting scenario. Based on the three-dimensional model of the uneven area, the surface morphology in the next 0.1 - 0.2 seconds is predicted, and the control instructions are calculated and cached in advance. The prediction time window satisfies the relation: ， Among them, is the control delay, is the system execution delay, is the safety factor; the feedforward control is automatically activated when the cutting speed v exceeds the critical value at that time.

7. The laser cutting method for splicing building steel plates according to claim 6, characterized in that In Step S4, the cutting process is monitored in real time through 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 score for the consistency of the slit width, T is the score for the rationality of the temperature distribution, A is the score for the regularity of the acoustic characteristics, R is the score for the stability of the reflected light intensity, and w1 to w4 are weight coefficients; when Q is lower than the threshold parameter fine-tuning is triggered, , where is the measure of unevenness in the local area, and G0 is the reference value; the amplitude of the parameter fine-tuning satisfies the condition: ≤0.15, ≤0.2, ≤0.3mm, where ΔP is the power adjustment amount, ΔV is the speed adjustment amount, V0 is the current cutting speed, and ΔZ is the height adjustment amount.

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

9. A cutting production line structure for implementing the laser cutting method for splicing building steel plates according to claim 1, characterized in that, Comprising: A preliminary detection module, including a laser irradiation unit and a preliminary deformation analysis unit, which is used to irradiate the joint area and analyze the dislocation deformation, judge 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 start and end 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 form data construction unit, which receives the uneven area range data, scans and constructs the three-dimensional form data of the uneven area; A dynamic cutting module, including a laser cutting unit and a cutting parameter adjustment unit, which adjusts the height, focal length and power of the laser cutting head according to the three-dimensional form data to achieve precise cutting of the uneven area; Among them, the detection result of the preliminary detection module triggers the length measurement module, determines the range of the uneven area and provides an accurate scanning area for the three-dimensional data acquisition module, and the collected data provides a basis for real-time cutting parameter adjustment for the dynamic cutting module, so as to achieve precise cutting control.

Images