An adaptive thin plate laser cutting quality closed-loop optimization method and device

Abstract

The embodiment of the application provides a kind of adaptive sheet laser cutting quality closed loop optimization method and device, it is related to laser cutting technical field, method includes scanning and establishing initial three-dimensional curved surface model and marking non-contact area by cutting head two sides array laser displacement sensor before cutting;Path pre-compensation is carried out based on multivariate regression thermal deformation empirical model, and adaptive micro-hole array is generated in the expected warping section to actively release internal stress;During cutting, the same axis optical sensor and array displacement sensor data are fused to monitor slit width and slag splashing in real time, and laser power, cutting speed and auxiliary air pressure are dynamically linked and adjusted;After cutting, the same sensor array verifies deviation and intelligently disposes;Recursive least squares (RLS) algorithm is used to close loop correct empirical model parameters.The application has the effect of reducing sheet flatness deviation, improving productivity and reducing scrap rate.

Description

Technical Field

[0001] This invention relates to the field of laser cutting technology, and in particular to an adaptive closed-loop optimization method and apparatus for thin plate laser cutting quality. Background Technology

[0002] Laser cutting is the mainstream process for cutting thin plate structural components such as battery brackets and battery swapping pile bases for new energy vehicles. It has advantages such as high speed, high precision and non-contact operation. However, for thin plates of 0.5–3mm, there is a fundamental contradiction between "high-speed cutting" and "low thermal deformation": high-power and high-speed cutting leads to concentrated heat input, resulting in significant thermal expansion of the plate, accumulation of residual stress and out-of-tolerance flatness. This requires a lot of manual leveling or scrapping, which seriously restricts production efficiency and product quality.

[0003] Existing technologies mainly employ post-processing remedial measures or single compensation methods. CN114918553B discloses an adaptive control method for thermal deformation in laser etching, which adjusts the processing sequence by real-time measurement of deformation through gridded regional scanning and machine vision, but does not involve pre-compensation for cutting paths or active stress release. CN117123938B discloses a power compensation method for thin-film laser cutting, which adjusts the power in real time by matching the spot contour, but is limited to adjusting a single power parameter, does not integrate multiple sensors, and lacks closed-loop model iteration. CN101376194A discloses an online measurement and compensation device for laser welding, which compensates for process parameters in real time by measuring the seam morphology in advance using a camera, but it is applied to welding rather than cutting and lacks the micropore stress release mechanism unique to thin plates.

[0004] To address the aforementioned issues, there is an urgent need for a method and device that can actively sense and release thermal stress during the cutting process and achieve closed-loop self-optimization of quality. Summary of the Invention

[0005] To address the aforementioned issues, this application provides an adaptive closed-loop optimization method and apparatus for thin-plate laser cutting quality.

[0006] This application provides an adaptive closed-loop optimization method and apparatus for thin-plate laser cutting quality, which adopts the following technical solution: An adaptive closed-loop optimization method for the quality of thin plate laser cutting includes the following steps: S1. Pre-cutting reference calibration and adaptive support prediction for non-contact areas: The sheet to be processed is placed on the worktable, and the initial surface contour is scanned by the array laser displacement sensors installed on both sides of the cutting head to establish a three-dimensional curved surface model M0(x,y). At the same time, the gap distribution between the sheet and the worktable is detected, the non-contact areas with gaps greater than a preset threshold (such as 0.2mm) are marked, and an adaptive support compensation signal is generated. S2. Thermal Deformation Pre-compensation Path Planning and Active Stress Release: Based on the thermal deformation empirical model, the thermal shrinkage amount ΔL(s) and expected warpage height H(s) at each position along the cutting path are estimated; the original design contour is geometrically pre-deformed and offset in reverse, and the offset amount is positively correlated with ΔL(s); in the section where the expected warpage height H(s) exceeds the threshold, an adaptive micropore array is automatically generated, and micropores are formed synchronously during the cutting process to achieve active release of internal stress; S3. Real-time monitoring with multi-sensor fusion during the cutting process: An optical sensor coaxial with the laser beam is used to monitor the kerf width W(t) and slag spatter characteristics in real time; at the same time, the real-time data of the array-type laser displacement sensor is fused to predict the thermal deformation deviation at the current cutting position; S4. Dynamic compensation control: Based on the monitoring signal of S3, the laser power P(t), cutting speed v(t), and auxiliary gas pressure Q(t) are adjusted in real time; when the non-contact area marked by S1 enters the cutting position, support compensation is triggered simultaneously. S5. Online verification and intelligent handling after cutting: After cutting, the cut contour and the surface of the board are re-scanned by the same array of laser displacement sensors. The scanning results are compared with the design CAD model to calculate the actual deviation δ(x,y) at each position. When δ(x,y)≤0.3mm, it is marked as qualified and recorded in the quality traceability database. When δ(x,y)>0.3mm, the local secondary cutting path is automatically planned or leveling parameter suggestions are generated. S6. Closed-loop self-learning optimization: The actual deviation δ(x,y) measured after each cut is compared with the predicted value in step S2. The parameters of the thermal deformation empirical model are corrected in reverse using the recursive least squares (RLS) algorithm, so that the prediction accuracy of the model is improved with each processing batch.

[0007] Optionally, the array of laser displacement sensors consists of 8–16 units, arranged at equal intervals on both sides of the cutting head, with a spacing of 50–100 mm.

[0008] Optionally, the reverse offset of the geometric pre-deformation is taken as α·ΔL(s), where α is the pre-deformation coefficient, with a value range of 0.6–0.9, and it decreases adaptively with the increase of plate thickness.

[0009] Optionally, the micropore diameter of the micropore array is 0.5–1 mm, the spacing between pores is 10–20 mm, and the pore depth is controlled to be 60%–80% of the plate thickness; the position and number of micropores are determined by the gradient algorithm of the expected warp height H(s): when dH(s) / ds exceeds the preset threshold, micropores are inserted in proportion to H(s), and the total area of ​​micropores does not exceed 2% of the area of ​​the cut contour.

[0010] Optionally, the weight factor λ of the recursive least squares (RLS) algorithm has a value range of 0.95–0.98.

[0011] The present invention also provides an apparatus for implementing an adaptive closed-loop optimization method for the quality of thin plate laser cutting, comprising a laser cutting head integrated module: including a laser cutting head body, a coaxial optical sensor installed inside the cutting head and coaxially arranged with the laser beam, and an array of laser displacement sensors installed on both sides of the cutting head; Control system: includes an industrial computer and a programmable logic controller (PLC). The industrial computer integrates a path planning module, a micro-hole array generation module, a dynamic parameter linkage control module, and a recursive least squares self-learning module. The PLC is used to receive sensor signals and execute control instructions. Non-contact area support module: An air cushion support unit set on the workbench, used to provide auxiliary support in the non-contact area marked S1.

[0012] Optionally, the device enables data communication and command interaction between modules via a bus or industrial Ethernet, is compatible with existing laser cutting equipment, and achieves closed-loop optimization by adding a sensor array.

[0013] In summary, this application includes at least one of the following beneficial technical effects: By using an adaptive micro-hole array to simultaneously release internal thermal stress during the cutting process, warping is reduced from the source. Compared with simple power compensation, the flatness deviation of thin plates can be reduced by more than 40%. The RLS algorithm is used to improve the model accuracy with batch iteration. The error of the first piece >0.5mm can be converged to <0.15mm within 3-5 pieces, which is particularly suitable for multi-variety mixed production lines. The flatness requirements are directly met after cutting, eliminating the need for manual leveling and improving the production line cycle time. Attached Figure Description

[0014] Figure 1 This is the overall flowchart of this method. Detailed Implementation

[0015] The terminology used in the following embodiments is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. As used in the specification and appended claims of this application, the singular expressions “a,” “an,” “the,” “the,” “the,” and “this” are intended to also include expressions such as “one or more,” unless the context clearly indicates otherwise. It should also be understood that in the following embodiments of this application, “at least one” and “one or more” refer to one, two, or more than two. The term “and / or” is used to describe the relationship between related objects, indicating that three relationships may exist; for example, A and / or B can indicate: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. The character “ / ” generally indicates that the preceding and following related objects are in an “or” relationship.

[0016] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0017] The following is in conjunction with the appendix Figure 1 The present invention will be described in further detail below.

[0018] This application discloses an adaptive closed-loop optimization method and apparatus for thin-plate laser cutting quality, referring to... Figure 1 This includes the following steps: S1. Pre-cutting reference calibration and adaptive support prediction for non-contact areas: Place the sheet to be processed on the worktable and scan the initial surface contour using array-type laser displacement sensors installed on both sides of the cutting head. Establish a three-dimensional surface model M0(x,y) with 8–16 measuring points spaced 50–100mm apart. Simultaneously detect the gap distribution between the sheet and the worktable, mark the non-contact areas with gaps >0.2mm, and generate adaptive support compensation signals.

[0019] S2. Thermal Deformation Pre-compensation Path Planning and Active Stress Release: The system utilizes a thermal deformation experience model specific to the sheet material, thickness, and cutting path to estimate the thermal shrinkage ΔL(s) and warpage height H(s) at each location. A geometric pre-deformation reverse offset is applied to the original design contour, with the offset amount = α × ΔL(s), where α is the pre-deformation coefficient, ranging from 0.6 to 0.9, and decreasing with increasing sheet thickness. In the expected warpage section where H(s) > 0.15 mm, an adaptive micro-hole array is automatically generated, with a diameter of 0.5–1 mm and a spacing of 10–20 mm. The location and number of micro-holes are determined by the H(s) gradient algorithm, and the hole depth is controlled to be 60–80% of the sheet thickness. The location and number of micro-holes are determined by the gradient of the expected warpage height, and the total area of ​​the micro-holes does not exceed 2% of the cutting contour area. These micro-holes are formed synchronously during the cutting process, achieving active release of internal stress.

[0020] The thermal deformation empirical model is a multivariate regression model based on historical processing data. Its input parameters include: sheet material: coefficient of thermal expansion α, thermal conductivity λ, specific heat capacity c, thickness t, laser power P, cutting speed v, auxiliary gas pressure Q, total cutting path length L, and local radius of curvature r; the outputs are the thermal shrinkage ΔL(s) and expected warpage height H(s) along the path parameter s. The basic form of the model is: ΔL(s)=β0+β1·(P / v)·α·t+β2·L(s)+β3·(1 / r(s))+ε, H(s)=γ0+γ1·(P / v)·t²+γ2·∫ΔT(s)ds+ε', Where βᵢ and γᵢ are regression coefficients obtained by fitting the historical cutting experimental data accumulated from the company's 12 laser cutting equipment, and ε is the residual term; for example, for 1.5mm aluminum alloy, under the conditions of P=3000W and v=80mm / s, the following values ​​were obtained by fitting 30 sets of experiments: β1≈0.025, β2≈0.005, β3≈0.6, γ1≈0.12, γ2≈0.008.

[0021] The initial model was established using the offline least squares method. In practical applications, the model parameters can be updated online through the recursive least squares (RLS) algorithm of S6 to achieve adaptive optimization. The generation algorithm of the micropore array is based on the H(s) gradient: when dH(s) / ds exceeds the preset threshold, micropores are inserted in this segment according to the proportion of H(s) value, with the number N=k·H(s), where k is an empirical coefficient of 0.8–1.2, to ensure that the internal stress is released without affecting the structural strength, that is, the total area of ​​micropores is less than 2% of the area of ​​the cut contour.

[0022] S3. Real-time monitoring with multi-sensor fusion during cutting: The coaxial optical sensor built into the cutting head is used to monitor the kerf width W(t) and slag splash characteristics in real time. When W(t) deviates from ±10% or the splash pattern changes abruptly, it is determined to be a local heat input abnormality. At the same time, the array laser displacement sensor monitors the surface height change of the plate near the cutting path at a frequency of 1kHz to predict the thermal deformation trend.

[0023] S4. Dynamic Compensation Control: Based on the S3 fusion signal, the laser power P(t), cutting speed v(t), and auxiliary air pressure Q(t) are adjusted in real time. When the non-contact area is triggered, support compensation is activated to further stabilize deformation. If the kerf is too wide, the laser power is reduced by 5%–15%, and the cutting speed is reduced proportionally. If the kerf is too narrow, the power and speed are increased. The auxiliary air pressure is automatically increased by 10%–20% when there is abnormal splashing. When cutting to the non-contact area marked by S1, the air cushion support unit on the worktable is activated to provide back pressure support of 0.2–0.4MPa to prevent secondary warping of the sheet metal.

[0024] S5. Online verification and intelligent handling after cutting: The cut contour and surface are rescanned with the same sensor array and compared with the CAD model to calculate the actual deviation δ(x,y); δ≤0.3mm is marked as qualified and added to the quality traceability database; δ>0.3mm automatically plans the local secondary micro-cutting path or generates leveling suggestions.

[0025] S6. Closed-loop self-learning optimization: The actual δ(x,y) is compared with the predicted value in S2. The recursive least squares (RLS) algorithm is used with a weight factor λ = 0.95–0.98 to correct the parameters of the thermal deformation empirical model in reverse, thereby achieving iterative improvement of the cutting accuracy of subsequent workpieces of the same type. The recursive formula of the RLS algorithm is as follows: K(k) = P(k-1)φ(k) / [λ + φᵀ(k)P(k-1)φ(k)] θ̂(k) = θ̂(k-1) + K(k)[y(k) - φᵀ(k)θ̂(k-1)] P(k) = [I - K(k)φᵀ(k)]P(k-1) / λ Where θ̂ is the coefficient vector to be estimated (βᵢ or γᵢ), φ is the input vector, y is the measured deviation, λ=0.96, and after 5 iterations, the prediction error of the same type of workpiece is reduced from 0.52mm to 0.12mm.

[0026] The present invention also provides an apparatus for implementing the above method, comprising a laser cutting head integrated module, a control system, and an optional non-contact area support module; the laser cutting head integrated module includes a laser cutting head body, a coaxial optical sensor installed inside the cutting head and coaxially arranged with the laser beam, and an array-type laser displacement sensor installed on both sides of the cutting head; the array-type laser displacement sensor is used to scan the initial contour of the plate before cutting, monitor real-time deformation during cutting, and scan the cut surface after cutting; the control system includes an industrial computer and a programmable logic controller (PLC); the industrial computer integrates a path planning software module, a micro-hole array generation algorithm module, a dynamic parameter linkage control module, and a recursive least squares self-learning module; the programmable logic controller is used to receive sensor signals, execute control commands, and drive the laser, servo motor, and auxiliary air circuit.

[0027] The non-contact area support module is an air cushion support unit set on the workbench, which is used to provide auxiliary support for the non-contact area marked in step S1 to prevent secondary warping of the board.

[0028] The above modules communicate and exchange commands via bus or industrial Ethernet to form a complete closed-loop control system.

[0029] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. Any equivalent modifications or variations made by those skilled in the art based on the content disclosed in the present invention should be included within the scope of protection set forth in the claims.

Claims

1. An adaptive closed-loop optimization method for the quality of thin plate laser cutting, characterized in that, Includes the following steps: S1. Before cutting, the initial surface contour is scanned by an array of laser displacement sensors installed on both sides of the cutting head to establish a three-dimensional curved surface model and mark the non-contact area between the plate and the worktable. S2. Based on the thermal deformation experience model, the thermal shrinkage and expected warpage height at each position along the cutting path are estimated. The original design contour is geometrically pre-deformed and offset in the opposite direction. Micro-hole arrays are automatically generated in the section where the expected warpage height exceeds the threshold. Micro-holes are formed synchronously during the cutting process to release internal stress. S3. During the cutting process, an optical sensor coaxial with the laser beam is used to monitor the kerf width and slag splash characteristics. At the same time, the real-time data of the array-type laser displacement sensor is fused to predict thermal deformation deviation. S4. Based on the monitoring signal of S3, adjust the laser power, cutting speed and auxiliary air pressure in real time, and trigger support compensation when cutting to the non-contact area marked by S1. S5. After the cutting is completed, the array-type laser displacement sensor rescans the cut contour and calculates the actual deviation. If the deviation exceeds the threshold, a secondary cutting or leveling path is automatically planned. S6. Compare the actual deviation with the predicted deviation in S2, and use the recursive least squares algorithm to correct the parameters of the thermal deformation empirical model in reverse.

2. The adaptive thin-plate laser cutting quality closed-loop optimization method according to claim 1, characterized in that: The array of laser displacement sensors consists of 8–16 units with a spacing of 50–100 mm.

3. The adaptive thin-plate laser cutting quality closed-loop optimization method according to claim 1, characterized in that: The reverse offset of the geometric pre-deformation is α·ΔL(s), where ΔL(s) is the estimated thermal shrinkage, and α is the pre-deformation coefficient, which ranges from 0.6 to 0.9 and decreases with the increase of plate thickness.

4. The adaptive thin-plate laser cutting quality closed-loop optimization method according to claim 1, characterized in that: The micropore array has a micropore diameter of 0.5–1 mm, a micropore spacing of 10–20 mm, and a micropore depth of 60%–80% of the plate thickness. The position and number of micropores are determined by the gradient of the expected warp height, and the total area of ​​the micropores does not exceed 2% of the area of ​​the cut contour.

5. An adaptive thin-plate laser cutting device, characterized in that: The method described by any one of claims 1-4 includes Laser cutting head integrated module: includes laser cutting head body, coaxial optical sensor installed inside the cutting head and coaxial with the laser beam, and array-type laser displacement sensors installed on both sides of the cutting head; Control system: includes an industrial computer and a programmable logic controller. The industrial computer integrates a path planning module, a micro-hole array generation module, a dynamic parameter linkage control module, and a recursive least squares self-learning module. Non-contact area support module: an air cushion support unit set on the workbench, used to provide auxiliary support in the non-contact area marked S1.

6. The apparatus according to claim 5, characterized in that: The device enables data communication and command interaction between modules via bus or industrial Ethernet, is compatible with existing laser cutting equipment, and achieves closed-loop optimization by adding a sensor array.

Images