Laser positioning stripping system and method for special-shaped NdFeB scrap

Abstract

The present invention relates to the technical field of material removal in laser processing, and in particular to a laser positioning stripping system and method for special-shaped NdFeB scrap. The method comprises: S1: performing multispectral scanning on the surface of the special-shaped NdFeB scrap; S2: dynamically adjusting the energy parameters of the laser stripping and constructing an energy-thickness matching model; S3: generating a fractal scanning path adaptive to the special-shaped surface based on the geometric profile data of the three-dimensional characteristic spectrum; S4: performing in-situ surface roughness detection during the laser stripping process, triggering a secondary refinement mode based on the detection results, and dynamically correcting the safe stripping margin based on the stripping depth data; S5: combining a hot air flow field with the principles of gas-solid separation dynamics to achieve in-situ separation of oxide layer fragments and matrix particles, and enhancing the recovery of matrix particles with the assistance of a magnetic field. The system can dynamically adjust the laser energy, scanning path, and thermal control strategy based on real-time data, thereby effectively avoiding material loss and waste and improving processing efficiency and recovery rate.

Description

Technical Field

[0001] The present invention relates to the technical field of material removal in laser processing, and in particular to a laser positioning and stripping system and method for special-shaped NdFeB waste. Background Art

[0002] Neodymium iron boron permanent magnets, as key functional materials in modern industry, are widely used in new energy vehicles, precision instruments, and aerospace. With the surge in the use of complex, irregularly shaped components, the recycling of geometrically irregular scrap materials generated during the production process (such as turbine engine magnetic ring fragments and irregular sensor cores) has become an industry challenge. Traditional pickling methods face serious limitations when treating such irregular scrap: strong acid solutions struggle to evenly penetrate complex curved surfaces, resulting in incomplete oxide layer stripping and excessive corrosion of the substrate. Rare earth element recovery rates are less than 75%, and large amounts of heavy metal-contaminated wastewater are generated. While mechanical stripping techniques can avoid chemical contamination, they are poorly adapted to thin-walled, curved structures and pose the risk of microcrack propagation in the substrate. Statistics show that the breakage rate of irregularly shaped parts less than 1 mm thick is as high as 34%, severely restricting the recycling of high-value waste.

[0003] While the recently emerging laser lift-off technology offers environmental advantages, it has exposed significant technical bottlenecks in the processing of shaped NdFeB scrap. First, conventional laser scanning systems rely on single-wavelength imaging and are unable to penetrate the oxide layer to accurately identify the substrate contour, resulting in positioning errors that trigger substrate ablation. Measurements show that the misprocessing rate in the curved transition zone exceeds 18%. Second, the sudden change in surface curvature of shaped parts exacerbates the thermal accumulation effect. Existing equal-energy scanning strategies are prone to uncontrolled material phase transformations in the groove area, causing grain boundary oxidation depths exceeding 5μm and severely damaging the magnetic properties of the substrate. Furthermore, traditional linear scanning paths have difficulty matching complex geometric features, resulting in processing blind spots of up to 27% in narrow slits with a curvature radius of less than 3mm. Furthermore, the sorting process of the lift-off products is physically isolated from the processing process, necessitating the introduction of additional vibration screening equipment. This results in a loss rate of rare earth oxides with a particle size of less than 50μm exceeding 12%, and secondary oxidation of the substrate particles is difficult to avoid. Summary of the Invention

[0004] In order to solve the technical problem of misprocessing of special-shaped NdFeB scrap, the present invention provides a laser positioning and stripping system and method for special-shaped NdFeB scrap.

[0005] The technical solutions provided by the embodiments of the present invention are as follows:

[0006] The laser positioning stripping method for irregularly shaped NdFeB scrap provided by an embodiment of the present invention comprises: S1: performing multispectral scanning on the surface of the irregularly shaped NdFeB scrap, obtaining geometric profile data of the substrate material and oxide layer thickness distribution data through synchronous scanning with a long-wavelength laser and a short-wavelength laser, and establishing a three-dimensional characteristic spectrum of the oxide layer-substrate boundary based on the difference characteristics of the dual-band reflection signals;

[0007] S2: dynamically adjusting the energy parameters of laser stripping based on the oxide layer thickness distribution data in the three-dimensional characteristic spectrum, constructing an energy-thickness matching model, implementing laser stripping using a step-by-step energy control strategy, and monitoring and terminating irradiation in real time through plasma spectroscopy when approaching the substrate interface;

[0008] S3: generating a fractal scanning path for an adaptive profiled surface based on the geometric profile data of the three-dimensional characteristic spectrum, and dynamically adjusting the scanning path and cooling interval according to real-time thermal accumulation data;

[0009] S4: Perform in-situ surface roughness testing during the laser lift-off process, trigger the secondary refinement mode based on the test results, and dynamically correct the safety lift-off margin based on the lift-off depth data;

[0010] S5: The thermal gas flow field generated during the laser stripping process is used in combination with the gas-solid separation dynamics principle to achieve in-situ separation of oxide layer fragments and matrix particles, and the recovery of matrix particles is enhanced by the assistance of a magnetic field.

[0011] Accordingly, the embodiment of the present invention further provides a laser positioning and stripping system for special-shaped NdFeB scrap, which is used to run the laser positioning and stripping method for special-shaped NdFeB scrap described in the embodiment of the present invention, including:

[0012] Multispectral scanning module, including:

[0013] a long wavelength laser scanning unit configured to emit near infrared laser light that penetrates the oxide layer and collects geometric profile data of the substrate;

[0014] a short-wavelength laser scanning unit configured to emit visible laser light that interacts with electron energy levels of the oxide layer and acquire optical response data of the oxide layer;

[0015] a signal fusion processing unit, electrically connected to the long-wavelength and short-wavelength laser scanning units, configured to perform phase separation processing on the dual-band reflection signals and generate a three-dimensional characteristic spectrum;

[0016] The dynamic energy control module is connected to the multi-spectral scanning module via a data bus and includes:

[0017] an energy calculation engine configured to receive oxide layer thickness data in a three-dimensional characteristic spectrum and calculate a stepped energy parameter;

[0018] A laser generating unit, electrically connected to the energy calculation engine, comprising an adjustable pulse width fiber laser and a beam shaping component;

[0019] Plasma monitoring unit, integrating fiber optic spectrometer and characteristic peak recognition circuit, outputs laser termination signal in real time;

[0020] The path planning control module is connected to the multi-spectral scanning module and the dynamic energy control module through a high-speed data interface, including:

[0021] a fractal path generator configured to generate a Hilbert curve and a spiral progressive path based on geometric data of a three-dimensional characteristic spectrum;

[0022] Thermal field simulation unit, with built-in material thermal property database and finite element calculation core, outputs heat accumulation warning signal;

[0023] A motion controller, mechanically connected to the laser scanning galvanometer, receives path instructions and drives the optical actuator;

[0024] Quality feedback correction module, including:

[0025] a confocal microscopy imaging unit optically coupled to the processing area and configured to collect surface topography data of a micro-area;

[0026] The roughness analysis processor is connected to the microscopic imaging unit through an image acquisition card and has a built-in three-dimensional Fourier transform algorithm library;

[0027] The safety margin calculation unit interacts with the dynamic energy control module through a data line and outputs a processing termination instruction;

[0028] The gas-solid separation and recovery module is physically connected to the process chamber outlet and includes:

[0029] High-temperature airflow generating device, integrated with temperature sensor and flow control valve, forms thermodynamic coupling with the thermal field of laser processing area;

[0030] The magnetoelectric composite separation channel has gradient magnetic field generators and electrode arrays arranged in sequence along the direction of material movement;

[0031] Particle monitoring components, including high-speed cameras and image processors, provide feedback to adjust sorting parameters;

[0032] The process optimization subsystem communicates with each module via industrial Ethernet, including:

[0033] A machine learning accelerator card configured to run a deep neural network model and output a process parameter optimization solution;

[0034] Digital twin server, which stores multi-physics field coupling simulation models and compares them with actual processing data in real time;

[0035] Parameter calibration interface, injecting compensation coefficients into the motion controller and laser generating unit;

[0036] The output end of the multi-spectral scanning module transmits the three-dimensional characteristic spectrum to the input end of the dynamic energy regulation module through optical fiber;

[0037] The path planning control module receives the geometric data from the multi-spectral scanning module and the thermodynamic parameters of the dynamic energy control module, and generates a control signal to drive the scanning galvanometer via the motion controller;

[0038] The optical signal input of the quality feedback correction module comes from the reflected light path of the processing area, and the correction instruction output by it is synchronously sent to the dynamic energy regulation module and the path planning control module through the digital I / O interface;

[0039] The airflow parameters of the gas-solid separation and recovery module are controlled by the real-time energy data of the dynamic energy control module, and its particle monitoring signal is fed back to the mass feedback correction module;

[0040] The process optimization subsystem establishes data channels with all modules through the OPC UA protocol to form a closed-loop control network.

[0041] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0042] Through multi-spectral scanning, precise energy control and intelligent path planning, efficient stripping of the oxide layer and the matrix is ​​achieved, minimizing matrix damage. Through real-time monitoring of the plasma spectrum, in-situ roughness detection and heat accumulation control, precise control of surface quality and stripping depth is ensured. Combining gas-solid sorting dynamics and magnetic field-assisted recovery technology, the present invention can efficiently sort oxide layer fragments and matrix particles, improving recovery efficiency. Through continuous optimization of machine learning and digital twin systems, process stability and production efficiency are improved, while production costs are reduced. The present invention has significant advantages in improving the recovery rate of NdFeB waste, optimizing processing quality and reducing energy consumption. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0044] Figure 1 A flowchart of the steps of the laser positioning stripping method for irregular-shaped NdFeB scrap provided by an embodiment of the present invention;

[0045] Figure 2A flowchart of step S1 in the laser positioning stripping method for irregular-shaped NdFeB scrap provided in an embodiment of the present invention;

[0046] Figure 3 This is a flow chart of step S2 in the laser positioning stripping method for special-shaped NdFeB scrap provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0047] The technical solutions of the present invention are described below with reference to the accompanying drawings. It is also noted that, to provide a more detailed description, the following embodiments are best and preferred embodiments, and those skilled in the art may employ alternative implementations for certain known technologies. Furthermore, the accompanying drawings are intended only to provide a more detailed description of the embodiments and are not intended to limit the present invention.

[0048] It should be noted that references in the specification to "one embodiment," "an embodiment," "exemplary embodiments," "some embodiments," etc. indicate that the described embodiments may include specific features, structures, or characteristics, but not necessarily every embodiment will include such specific features, structures, or characteristics. Furthermore, when specific features, structures, or characteristics are described in conjunction with an embodiment, it is within the knowledge of persons skilled in the relevant art to implement such features, structures, or characteristics in conjunction with other embodiments (whether or not explicitly described).

[0049] In general, terms can be understood, at least in part, from their use in context. For example, depending at least in part on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending at least in part on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0050] It will be understood that the meanings of “on,” “over,” and “above” in the present invention should be interpreted in the broadest manner, so that “on” means not only “directly on” something but also includes the meaning of being “on” something with intervening features or layers, and “on” or “above” means not only “on” or “above” something but also includes the meaning of being “on” or “above” something with no intervening features or layers.

[0051] Additionally, spatially relative terms such as "below," "beneath," "lower," "above," and "upper" may be used herein for descriptive convenience to describe the relationship of one element or feature to another element or features, as illustrated in the accompanying drawings. Spatially relative terms are intended to encompass different orientations of the device in use or operation in addition to the orientation depicted in the accompanying drawings. The device may be oriented in other ways, and the spatially relative descriptors used herein should be similarly interpreted accordingly.

[0052] like Figures 1 to 3 As shown, an embodiment of the present invention provides a laser positioning stripping method for special-shaped NdFeB scrap. In step S1, the scrap surface is first scanned synchronously by a long-wavelength laser and a short-wavelength laser to obtain the geometric contour data of the matrix material and the thickness distribution data of the oxide layer. The long-wavelength laser can penetrate the oxide layer to obtain the geometric morphology information of the matrix, while the short-wavelength laser interacts with the electronic structure of the oxide layer to obtain the thickness distribution data of the oxide layer. Based on the difference characteristics of the dual-band reflection signal, a three-dimensional characteristic spectrum of the oxide layer and the matrix is ​​constructed to provide accurate data support for the subsequent stripping process. This step provides accurate three-dimensional geometric data and oxide layer thickness distribution for the subsequent laser stripping operation, ensuring the efficiency and accuracy of laser stripping.

[0053] Based on the oxide layer thickness data acquired in step S1, the laser stripping energy parameters are dynamically adjusted, and a stepped energy control strategy is implemented by constructing an energy-thickness matching model. As the laser approaches the substrate interface, the energy of the laser stripping process is automatically reduced, and the oxide layer stripping progress is monitored in real time using the plasma spectrum. If an anomaly in the plasma spectrum (e.g., an impact on the substrate material) is detected, laser irradiation is immediately terminated. This control mechanism effectively prevents damage to the substrate caused by excessive laser energy, protects the substrate material, reduces waste loss, and improves stripping accuracy.

[0054] In step S3, an adaptive fractal scanning path is generated based on the geometric profile data provided in step S1 to cover the scrap surface. The scanning path and cooling interval are dynamically adjusted based on real-time heat accumulation data. This step is particularly important because it prevents localized overheating that can damage the substrate or excessive oxide layer removal. By optimizing the scanning path and cooling interval, the temperature distribution can be effectively controlled, preventing overheating from affecting material quality and ensuring the stability and efficiency of the laser lift-off process.

[0055] During the laser lift-off process, in-situ surface roughness measurement technology monitors surface morphology in real time. When the surface roughness exceeds a set threshold, a secondary refinement mode is triggered, adjusting parameters such as laser power and scanning speed. The safe lift-off margin is dynamically corrected based on the lift-off depth to ensure a smooth surface that meets requirements. In-situ roughness measurement can promptly identify surface treatment anomalies, avoiding excessive finishing work later, ensuring surface quality meets standards, and improving process accuracy and efficiency.

[0056] During the laser stripping process, the thermal flow field generated by the laser helps separate the oxide layer fragments from the matrix particles. Utilizing the principles of gas-solid separation dynamics, the oxide layer fragments and matrix particles can be separated in situ. Simultaneously, the magnetic field-assisted recovery system enhances the recovery efficiency of the matrix particles, preventing the loss of valuable NdFeB material. The synergistic effect of the thermal flow field and magnetic field efficiently separates the oxide layer fragments from the matrix particles, improving not only the recovery rate but also the separation efficiency, thereby maximizing resource recovery.

[0057] Through the integrated approach described above, this laser-positioned stripping method achieves efficient and precise processing of irregularly shaped NdFeB scrap. Precise laser energy control, optimized scanning path adjustment, real-time roughness detection and trimming, and a thermal airflow and magnetic field-assisted sorting system all contribute to efficient material recovery and precise stripping. This method not only improves resource recovery, but also ensures the integrity of the substrate material and reduces waste loss, promising promising industrial applications.

[0058] In one possible implementation, long-wavelength laser scanning (S1.1) is first used to acquire geometric information about the substrate that can penetrate the oxide layer, generating raw point cloud data. Simultaneously, short-wavelength laser scanning (S1.2) captures the optical response of the oxide layer. Its reflected signal is sensitive to the oxide layer thickness, generating optical attenuation data. This synchronized scanning strategy simultaneously captures two key pieces of information in a single scan: the three-dimensional deformation characteristics of the substrate and the variation in oxide layer thickness. This synchronization improves data acquisition efficiency, avoids errors caused by repeated scanning, ensures a one-to-one correspondence between geometric information and optical properties, and achieves high-precision data coupling.

[0059] Furthermore, the two reflected signals are fed into a phase separation model, where their differential characteristics are analyzed in the time domain (e.g., phase delay) and frequency domain (e.g., energy distribution). This allows for refined interface identification factors, such as the reflection intensity ratio and phase difference. By leveraging phase and frequency domain information to enhance signal recognition, the boundary between the oxide layer and the substrate can be more accurately distinguished, particularly in areas with complex morphologies and minute thickness differences.

[0060] Based on these differential features, the scanned area is divided into subregions and feature vectors are extracted. Principal component analysis (PCA) is then used to simplify the data dimensions and extract key features. A dynamic classification model is then constructed using the support vector machine (SVM) algorithm to generate boundary recognition criteria that adapt to various morphological changes. Dynamic classification boundaries can adapt to changes in material properties in different regions, avoiding misjudgments caused by static thresholds and significantly improving recognition accuracy and robustness.

[0061] Based on this precise identification, a Poisson reconstruction algorithm is used to generate a geometric model of the substrate. The optical response data is then mapped onto the surface to construct an oxide layer thickness field. Finally, non-uniform rational B-splines (NURBS) are used for surface fitting to form a continuous, smooth three-dimensional oxide layer characteristic spectrum. NURBS fitting offers high smoothness and local adjustability, accurately reflecting subtle fluctuations in oxide layer thickness despite complex morphologies, providing a precise foundational model for subsequent laser energy control and path generation.

[0062] Each of these steps is tightly linked through layered data extraction and coupling, from initial multi-source scanning to final three-dimensional signature spectrum generation, forming a multimodal information fusion system for complex and irregular waste materials. This system offers significant advantages in high-contrast interface recognition, intelligent boundary classification, and continuous characterization, significantly improving the accuracy of subsequent laser stripping and the intelligence level of material identification. This method is suitable for recycling high-value, difficult-to-separate rare earth waste, combining efficiency, scalability, and industrial practicality.

[0063] In one possible implementation, the S2 stage primarily involves energy control and plasma spectrum monitoring during the laser lift-off process. This method utilizes a rationally designed energy-thickness matching model, a step-by-step energy control strategy, and a real-time plasma spectrum monitoring system to ensure precise control of the oxide layer and substrate during laser lift-off, minimizing material damage while simultaneously improving waste recovery efficiency.

[0064] Specifically, in step S2.1, an energy-thickness matching model is first constructed, describing the relationship between laser energy and oxide layer thickness through an exponential function. The exponential term coefficient of this model is determined by the material's light-to-heat conversion efficiency. This allows the laser energy to be adjusted according to the characteristics of different materials, ensuring that the laser heat is concentrated in the appropriate area while avoiding excessive energy that may damage the substrate. The model also needs to consider both the oxide layer's ablation threshold and the substrate damage threshold, forming a dual constraint to ensure that energy control at each stage can precisely adapt to the needs of different levels.

[0065] Through the precise design of the energy matching model, we can ensure effective laser stripping of the oxide layer while avoiding damage to the substrate caused by excessive energy. This model can dynamically adjust the laser parameters according to the specific conditions of the scrap, significantly improving stripping efficiency and material utilization.

[0066] In step S2.2, a stepped energy control strategy is used to adjust the laser output power in stages to account for the different material properties of the oxide layer and the substrate:

[0067] In the first stage, high-energy pulses are used to break through the dense oxide layer. The pulse width is selected to meet the pressure threshold required for the impact phase transition. The high-energy pulses in this stage can effectively break through thicker oxide layers.

[0068] In the second stage, the process switches to low-energy, long-pulse mode to treat the transition layer. During this stage, the low-energy pulse parameters are set to ensure that the depth of the heat-affected zone is less than the thickness of the transition zone between the oxide layer and the substrate, thus avoiding overheating the substrate and ensuring that the oxide layer is completely stripped.

[0069] In the third stage, when the laser approaches the matrix interface, the plasma emission spectrum is collected in real time to monitor the appearance of characteristic peaks of the matrix material, and timely determine whether it has reached the matrix interface to avoid excessive damage.

[0070] The stepped energy control strategy precisely controls the processing intensity of different material layers by adjusting the laser power in stages, avoiding uncontrolled energy transitions between layers. This approach significantly improves the accuracy and safety of laser lift-off, which is particularly important for the efficient processing of complex scrap materials.

[0071] In step S2.3, a fiber optic spectrometer collects the plasma emission spectrum within the 400-700nm wavelength range. Wavelet noise reduction technology is used to process the spectral data and extract the relative intensity ratios of the characteristic peaks of the rare earth elements. When the intensity ratio of the characteristic peaks exceeds a preset threshold, the laser output is immediately cut off. This real-time monitoring system accurately identifies the characteristic peaks of the matrix material and prevents laser damage to the substrate.

[0072] Real-time monitoring of the plasma spectrum ensures the dynamic adaptability of the laser lift-off process. It can detect the emergence of substrate material features in real time and then promptly cut off the laser output to avoid damage to the substrate caused by excessive treatment. This method significantly improves the safety and accuracy of the lift-off process and effectively protects the precious material.

[0073] These three steps, combined with an energy-thickness matching model, a step-by-step energy control strategy, and real-time plasma spectrum monitoring, form an efficient and safe laser stripping process. The energy-thickness matching model provides theoretical support, ensuring precise energy control; the step-by-step energy control strategy ensures layer-by-layer control during the actual stripping process, avoiding excessive energy input; and plasma spectrum monitoring provides real-time feedback at critical moments to prevent unnecessary damage to the substrate. Overall, this combination of three improves the accuracy, efficiency, and material recovery rate of the waste recovery process, resulting in a laser stripping method that combines both efficiency and safety.

[0074] In one possible implementation, the S3 stage primarily involves optimizing the laser scanning path and real-time control of heat accumulation. By utilizing a fractal scanning path generation method, real-time heat accumulation control, and cooling interval determination, the accuracy, efficiency, and material safety of laser lift-off can be effectively improved.

[0075] Specifically, in step S3.1, the processing area is first decomposed into multiple substructures, and appropriate filling patterns are selected based on the different local curvature characteristics. Specifically, when the curvature radius is less than a certain critical value, a spiral progressive path is used. This path allows for detailed laser processing in areas with less curvature, maximizing the material removal effect. For areas with greater curvature, a space-filling curve is used to optimize the path coverage density and ensure the efficiency of the removal process.

[0076] A recursive subdivision algorithm is implemented for high-complexity areas to further improve the adaptability of the path. The recursive subdivision criteria include the rate of change of curvature, the gradient of the oxide thickness, and the predicted heat accumulation. These factors combine to determine the optimization method of the path, effectively improving the accuracy of laser lift-off and avoiding improper treatment of high-complexity areas.

[0077] By selecting a path pattern tailored to the curvature characteristics, the accuracy and efficiency of the laser lift-off process can be maximized. The fractal scanning path method can adapt to scrap materials of varying shapes and complexities, ensuring more flexible and efficient laser processing and reducing material waste.

[0078] In step S3.2, a heat flux field prediction model is established, combining scanning path parameters, the material's thermal diffusivity, and laser energy parameters to calculate the temperature field distribution. This model can predict heat accumulation in local areas, ensuring that overheating and material damage are avoided during laser lift-off.

[0079] When the temperature gradient in a local area exceeds the set critical value, the following operations are performed:

[0080] Interrupt the current scan path and insert a cooling time interval: This operation pauses the laser scan and inserts a cooling time, allowing the temperature of the local area to decrease, thus avoiding excessive temperature that may damage the oxide layer or degrade the base material.

[0081] Switching to an adjacent cooler area to continue processing: This strategy allows laser processing to continue without affecting the overheated area.

[0082] Dynamically adjust the fill density of subsequent scanning paths: Based on real-time feedback of heat accumulation, dynamically adjust the fill density of subsequent paths to reduce heat accumulation and improve the overall thermal control effect.

[0083] Real-time heat accumulation control effectively prevents material damage caused by excessive laser energy or heat accumulation. This method ensures safety and efficiency during the laser lift-off process through precise temperature control, improving the quality of waste recycling.

[0084] In step S3.3, the heat dissipation time is calculated based on the heat diffusion equation. The cooling interval is then determined based on the material's characteristics and heat diffusion. Furthermore, the path sequence is replanned based on the priority of the subsequent processing areas. This ensures cooling while maximizing the priority of the processing areas, making the entire process more efficient.

[0085] By accurately calculating the heat dissipation time, heat accumulation can be effectively controlled. Re-planning the path sequence makes the insertion of cooling intervals more scientific and reasonable, avoiding ineffective waiting time and improving overall processing efficiency.

[0086] In one possible implementation, the S4 stage primarily involves detecting surface roughness, triggering secondary refinement mode, and dynamically adjusting the safe lift-off margin. These three steps optimize the accuracy and safety of the laser lift-off process, ensuring the required surface quality and avoiding unnecessary damage.

[0087] Specifically, in step S4.1, confocal laser scanning technology is first used to perform optical tomography along the normal direction of the peeled area to obtain micro-area surface morphology data. This method can accurately obtain the microscopic morphology of the material surface, providing detailed data for subsequent processing.

[0088] Furthermore, Fourier transform analysis is performed on the tomographic images to extract spatial frequency distribution characteristics, specifically the energy proportion of high-frequency components and the directional intensity distribution of medium-frequency components. These characteristics can be used to quantitatively analyze surface roughness and provide a basis for triggering subsequent refinement modes.

[0089] Furthermore, a surface roughness evaluation function is derived by weighting the deviation between the spatial frequency characteristics and the preset reference surface. This function can comprehensively evaluate the surface roughness and provide a standard for triggering the secondary refinement mode.

[0090] This method uses high-precision in-situ surface detection to accurately assess surface roughness, providing data support for subsequent refinement. Fourier transform analysis can better capture the fine features of the material surface, thereby improving the accuracy of the laser lift-off process.

[0091] Step S4.2 triggers the secondary finishing mode by setting a critical threshold of the surface roughness evaluation function. When the surface roughness exceeds the first critical threshold, the finishing mode is activated to further improve the surface quality.

[0092] During the finishing process, the system optimizes the processing effect by adjusting the following parameters:

[0093] Reduce laser power density: Reduce the power density to a fixed ratio of the original parameters to reduce the impact of excessive laser irradiation on the surface and avoid overheating or excessive peeling of the material.

[0094] Increase scanning speed: Increase the scanning speed to a fixed multiple of the original speed to reduce long-term exposure to the surface, thereby reducing thermal effects.

[0095] Reduce the spot overlap rate: Reduce the spot overlap rate to a preset safe range to ensure uniform distribution of laser energy and avoid unnecessary surface damage.

[0096] Monitoring during finishing: During the finishing process, the rate of change of surface roughness is continuously monitored. When the roughness change approaches zero, the finishing mode is automatically exited to ensure the efficiency and accuracy of the processing process.

[0097] By triggering the secondary finishing mode and precisely adjusting the laser parameters, the surface roughness can be effectively reduced to achieve the desired surface quality. At the same time, real-time monitoring during the finishing process ensures dynamic optimization of the machining process, helping to avoid unnecessary over-machining and material waste.

[0098] In step S4.3, a model for predicting the peel depth is established to predict the peel depth based on parameters such as the cumulative laser energy density, the thermal diffusivity of the material, and the porosity of the oxide layer. This model can calculate the ratio of the remaining substrate thickness to the original thickness in real time.

[0099] When the ratio falls below the second critical threshold, the system takes the following actions:

[0100] Immediately terminate laser irradiation in the current area: Stop laser irradiation in the current area to prevent material damage caused by excessive peeling.

[0101] Mark the forbidden processing area: mark the area as the forbidden processing area in the three-dimensional feature spectrum to ensure that the laser is no longer irradiated in this area.

[0102] Replanning path density: Replan the scan path density of the surrounding area to ensure the continuity and efficiency of the processing process, while avoiding unsafe areas from being processed again.

[0103] This method ensures safety during the laser lift-off process by predicting the lift-off depth in real time and correcting for the required margin, thus avoiding material loss or unnecessary risk caused by over-processing. Dynamic correction adjusts the scanning path based on actual conditions, making the laser lift-off process more accurate and safer.

[0104] The three steps of stage S4 work together to form a closed-loop control system. In-situ surface roughness measurement provides high-precision surface data, which serves as the basis for triggering the secondary finishing mode. Parameter adjustments in this finishing mode effectively improve surface quality. Simultaneously, a dynamic correction method for the safe stripping allowance ensures safe material processing by monitoring the stripping depth in real time. This combination of measures not only improves the accuracy and efficiency of laser lift-off, but also avoids damage caused by over-processing, ensuring the quality and safety of the material during processing.

[0105] In one possible implementation, step S5 involves gas-solid separation dynamics, magnetic field-assisted recovery methods, and closed-loop control of the separation process. These three technologies work together to improve waste separation efficiency, ensuring the effective recovery of useful matrix particles while removing oxide layer fragments.

[0106] Specifically, step S5.1 utilizes the high-temperature gas generated during the laser ablation process to construct a vertical airflow field. These high-temperature gases can adjust the turbulence intensity of the fluidized bed by controlling the airflow velocity, thereby affecting the particle settling process.

[0107] The design of the grading settling chamber is based on the difference in Stokes numbers between oxide layer fragments and matrix particles. The smaller particle size of the oxide layer fragments, while the higher mass density of the matrix particles, results in different settling velocities in the airflow. Therefore, a grading settling chamber is designed, with multiple stages of baffles to further refine the separation. The spacing between the baffles is determined based on the terminal velocity distribution of the particles, thereby improving separation accuracy.

[0108] The airflow adjustment and the design of the grading settling chamber can accurately separate the oxide layer fragments and the matrix particles, ensuring that the oxide layer fragments are effectively removed while retaining the matrix particles to the maximum extent. This not only improves the sorting efficiency but also reduces material waste.

[0109] In step S5.2, the system deploys a gradient magnetic field generator at the end of the separation process. The magnetic field strength gradually increases along the particle's direction of motion. This design enhances the attraction to the matrix particles and facilitates their recovery.

[0110] By pre-magnetizing the matrix particles, a residual magnetic moment is generated in the particles in an alternating magnetic field, enhancing their response to the magnetic field. Based on the Lorentz force principle, a particle trajectory control channel is designed. Its radius of curvature is inversely proportional to the magnetic field gradient, enabling precise control of the particle trajectory and effective recovery of the matrix particles.

[0111] By enhancing the magnetism of the base particles and controlling their trajectory, magnetic field-assisted recovery can effectively improve the recovery rate of base particles and avoid the accidental recovery of oxide layer fragments. The application of this technology further improves the selectivity and efficiency of the recovery process.

[0112] In step S5.3, the movement trajectory of the particles is monitored in real time by using a high-speed camera system. These camera systems can accurately capture the movement information of the particles and calculate the escape rate of the target particles through image recognition algorithms.

[0113] Based on the escape rate data feedback, the system dynamically adjusts the airflow speed and magnetic field strength, adjusting operating parameters in real time to minimize the escape rate. This ensures that more matrix particles are effectively sorted and recovered, and reduces the escape of ineffective particles.

[0114] Closed-loop control adjusts operating conditions in real time to ensure the optimization of the sorting process. By continuously monitoring the particle trajectory and dynamically adjusting the airflow and magnetic field strength, particles are prevented from escaping, improving sorting efficiency and ensuring the accuracy of the recycling process.

[0115] The three technical sub-steps in step S5 work together to achieve efficient waste sorting and recovery. Gas-solid separation dynamics provide initial separation for particle sorting, performing graded sedimentation based on the physical properties of the particles, effectively removing oxide layer fragments. Next, magnetic field-assisted recovery further improves the recovery rate of the matrix particles by enhancing their magnetic properties. Finally, closed-loop control of the sorting process ensures efficient operation of the entire sorting process through real-time monitoring and dynamic adjustment, minimizing particle escape.

[0116] The combination of these steps significantly improves the accuracy and efficiency of NdFeB scrap recycling, ensuring a high recovery rate of matrix particles and effectively reducing the loss of oxide layer fragments. The overall system optimizes every step of scrap sorting and provides an efficient and intelligent recycling solution.

[0117] In one possible implementation, the 3D Fourier transform analysis step in S4.1 is designed to extract material surface features through frequency-domain analysis of the image sequence, enabling efficient surface peeling quality assessment. This process includes conversion from the spatial domain to the frequency domain, extraction of spatial frequency features, and construction of a surface roughness evaluation function.

[0118] Specifically, in step S4.1.1, the purpose of this step is to convert the tomographic image sequence from the spatial domain to the frequency domain, so that spectral analysis can reveal the periodic and aperiodic characteristics of the material surface. First, a two-dimensional discrete Fourier transform (2D FFT) is performed on the image sequence along the XY plane to capture the frequency information of the material surface in the XY plane. Next, a one-dimensional Fourier transform (1D FFT) is performed along the Z axis to convert this tomographic image data into a three-dimensional spectrum.

[0119] The conversion from the spatial domain to the frequency domain provides a global frequency domain perspective, capable of capturing variations in surface features across all directions and scales. When processing irregularly shaped NdFeB scrap, the surface may exhibit complex morphologies and structures. Frequency domain conversion effectively reveals surface features of varying sizes, providing foundational data for subsequent frequency feature extraction and surface roughness assessment.

[0120] In step S4.1.2, the three-dimensional spectrum is divided into low-frequency, mid-frequency, and high-frequency regions. The boundary frequencies of each frequency region are determined by the physical properties of the material surface. Next, the integrated energy of the complex modulus within each frequency band is calculated and normalized to the total energy contribution. Furthermore, a directional filter is performed on the mid-frequency spectrum, and the energy distribution differences in the four directions of 0°, 45°, 90°, and 135° are calculated.

[0121] Frequency region division and energy distribution analysis effectively reveal the directional characteristics of surface roughness and texture. Low-frequency regions are typically associated with large-scale surface features (such as surface flatness and long-wavelength texture), while high-frequency regions reflect fine surface defects and microstructures. Directional filtering of the mid-frequency region can further analyze surface details, particularly directional characteristics, which is crucial for optimizing the focus and scanning direction of the laser beam during laser lift-off.

[0122] In step S4.1.3, a surface roughness evaluation function is constructed. First, a theoretical frequency distribution curve of an ideal peeling surface is defined as a reference surface. Next, the root mean square error (RMSE) between the actual surface frequency distribution and the reference curve is calculated at different frequency bands. The errors in each frequency band are then weighted and combined, with the weighting coefficients being related to the surface functional requirements (such as smoothness and peeling performance).

[0123] This process provides a basis for quantitatively evaluating surface quality by comparing the actual surface with the ideal one. Because different applications may require varying surface roughness and texture directionality, the weighting coefficients can be tailored to specific needs, ensuring precise control of the lift-off process. By weighting and superimposing these errors, a comprehensive evaluation function is ultimately derived, helping to optimize laser lift-off parameter settings.

[0124] These three steps are tightly integrated to form a complete surface quality assessment process. First, a 3D Fourier transform provides frequency domain data. Then, frequency binning and energy statistics are used to extract surface features. Finally, an evaluation function is constructed to quantitatively assess the actual surface roughness. The advantage of this design is that it can precisely optimize the laser lift-off process based on the complex surface characteristics of the material (such as roughness and texture in different directions), improving recycling efficiency while also controlling the post-lift-off surface quality.

[0125] Through this frequency analysis method, the surface quality of special-shaped NdFeB scrap can be efficiently evaluated, the processing requirements of scraps of different shapes can be adapted, the flexibility and accuracy of the laser positioning stripping method can be improved, and the high quality of the final recycled materials can be ensured.

[0126] In one possible embodiment, the design method of the graded settling chamber in S5.1 aims to effectively separate the oxide layer fragments and matrix particles and improve the recovery rate through precise airflow dynamics control and optimized analysis of particle behavior.

[0127] Specifically, the purpose of S5.1.1 is to optimize the relative motion of the airflow and particles by calculating the airflow velocity and the terminal velocity of the particles, ensuring effective separation of the oxide layer fragments from the matrix particles. First, the dynamic viscosity coefficient of the gas is calculated based on the gas temperature. Next, Stokes' law is used to derive the terminal velocity of the particles, and the influence of the particle shape factor is taken into account to more accurately calculate the terminal velocity of different particles. Finally, through iterative calculations, the airflow velocity that maximizes the velocity difference between the oxide layer fragments and the matrix particles is determined.

[0128] By precisely controlling the airflow velocity, you can ensure optimal particle separation in the sedimentation chamber. A reasonable airflow velocity can effectively improve the separation efficiency between debris and matrix particles, reduce mixing and recovery losses between particles, and provide a basis for the optimization of subsequent graded sedimentation.

[0129] The design of S5.1.2 focuses on optimizing particle distribution and collision behavior through baffle layout. First, a differential equation for particle trajectory is established, incorporating influencing factors such as particle mass, gas flow rate, and baffle geometry to accurately simulate particle trajectories. Then, a Monte Carlo simulation method is used to calculate the probability distribution of particle collisions and determine the likelihood of particle collisions under different baffle layouts. Finally, a genetic algorithm is used to optimize the target particle recovery rate to determine the optimal solution for baffle spacing.

[0130] The optimized layout of the multi-stage baffles effectively guides particle movement, controls particle settling velocity and distribution, and thus improves particle classification. The optimized baffle spacing ensures maximum particle recovery, reduces the loss of valuable materials in the waste, and improves recovery efficiency.

[0131] Furthermore, turbulence control is a key factor in ensuring stable particle separation during the settling process. A porous media rectifier layer at the settling chamber entrance effectively stabilizes the airflow and reduces instability caused by turbulence. Real-time monitoring of pressure fluctuations and adjustment of the air supply valve opening using a PID controller further precisely control airflow stability. When the Reynolds number exceeds a critical value, a turbulence suppression device is activated to prevent excessive turbulence from interfering with the particle settling process.

[0132] Controlling turbulence intensity maintains uniformity and stability of the airflow within the settling chamber, preventing excessive disturbances and ensuring accurate particle classification. The PID controller allows for fine-tuning of the airflow based on real-time data, further improving classification and particle recovery.

[0133] The design approach in S5.1 integrates fluidized bed dynamics, multi-stage baffle layout optimization, and turbulence control strategies to form an efficient graded sedimentation system. Each step is interconnected to ensure efficient particle separation during the sedimentation process. Fluidized bed dynamics parameter determination provides the basis for precise control of airflow velocity, multi-stage baffle optimization further improves particle recovery, and turbulence intensity control ensures the stability of the sedimentation process. The overall design has the beneficial effect of improving the accuracy and efficiency of waste sorting, reducing losses during the recycling process, optimizing the NdFeB waste recycling process, and maximizing the recovery value.

[0134] In one possible implementation, part S6 involves the establishment and continuous optimization of a process knowledge base, aiming to continuously improve processing efficiency and accuracy through machine learning, digital twins, and self-optimization mechanisms.

[0135] Specifically, the purpose of S6.1 is to optimize process parameters using machine learning technology by collecting and analyzing historical processing data. First, by collecting historical processing data (including three-dimensional characteristic spectrum parameters, laser energy sequence, surface roughness detection results, and sorting efficiency indicators), a training set is constructed to provide sufficient data support for the model. Next, a feature vector is defined, which includes important indicators such as the mean value of the oxide layer thickness gradient, the maximum heat accumulation value, and the path turning point density. These features can reflect the different states of the material during the laser stripping process. A deep neural network model is used for multi-objective optimization to output the optimal process parameter combination.

[0136] By training the machine learning model, we can find the optimal combination of process parameters for different processing conditions based on historical data. This method can continuously optimize the process as data accumulates, improving stripping efficiency, quality, and recovery rate.

[0137] Furthermore, the digital twin method verifies the accuracy of the actual processing process through physical field simulation in a virtual environment. First, a virtual processing environment is established and multiple physical field simulation modules are integrated, including a thermodynamic model of laser-material interaction, a molecular dynamics model of the oxide layer stripping process, and a computational fluid dynamics model of gas-solid two-phase flow. These models can simulate various physical phenomena in the laser stripping process. Then, the actual processing parameters are input into the digital twin system for multi-physics field coupling simulation. By comparing with the actual processing process, the deviation between the simulation results and the measured data is checked. If the deviation exceeds the allowable range, the system will trigger the model parameter calibration process and automatically adjust the model parameters to ensure more accurate simulation results.

[0138] Digital twin technology ensures a high degree of consistency between the virtual and actual machining processes, enabling simulation to verify the feasibility of optimization solutions and reducing the cost and time of the experimental phase. Furthermore, a dynamic calibration mechanism ensures continuous improvement of the machining process, ensuring that process parameters are always optimized.

[0139] S6.3 focuses on the continuous monitoring and optimization of process parameters during the production process. First, actual production data is regularly compared and analyzed with historical data in the knowledge base to identify drift patterns in process parameters. This comparison allows for timely detection of anomalies or changes in the production process. Next, appropriate compensation coefficients are generated and process parameters are adjusted to compensate for deviations. Optimization results are automatically fed into the control system through online updates, enabling real-time process optimization.

[0140] The self-optimization mechanism of process parameters can automatically adjust to changes in production in real time, ensuring the stability and efficiency of the production process. This not only improves the intelligence level of the production line, but also ensures stable and consistent product quality in the event of parameter fluctuations or environmental changes.

[0141] The S6 system combines machine learning, digital twins, and a self-optimization mechanism for process parameters to establish a closed-loop system for dynamic learning and real-time optimization. First, the machine learning model is trained using historical data to provide optimal process parameters. Then, digital twin technology verifies the process plan through multi-physics simulation, ensuring consistency between the design and the actual process. Finally, the self-optimization mechanism for process parameters ensures long-term stability and optimization of the process through real-time monitoring and adjustment. The overall design has the beneficial effects of improving production efficiency and product quality, reducing costs, and making the production process more intelligent, adaptable, and sustainable.

[0142] Accordingly, the embodiment of the present invention further provides 10. a laser positioning and stripping system for irregularly shaped NdFeB scrap, which is used to execute any of the laser positioning and stripping methods for irregularly shaped NdFeB scrap described in any embodiment of the present invention, comprising:

[0143] Multispectral scanning module, including:

[0144] a long wavelength laser scanning unit configured to emit near infrared laser light that penetrates the oxide layer and collects geometric profile data of the substrate;

[0145] a short-wavelength laser scanning unit configured to emit visible laser light that interacts with electron energy levels of the oxide layer and acquire optical response data of the oxide layer;

[0146] a signal fusion processing unit, electrically connected to the long-wavelength and short-wavelength laser scanning units, configured to perform phase separation processing on the dual-band reflection signals and generate a three-dimensional characteristic spectrum;

[0147] The dynamic energy control module is connected to the multi-spectral scanning module via a data bus and includes:

[0148] an energy calculation engine configured to receive oxide layer thickness data in a three-dimensional characteristic spectrum and calculate a stepped energy parameter;

[0149] A laser generating unit, electrically connected to the energy calculation engine, comprising an adjustable pulse width fiber laser and a beam shaping component;

[0150] Plasma monitoring unit, integrating fiber optic spectrometer and characteristic peak recognition circuit, outputs laser termination signal in real time;

[0151] The path planning control module is connected to the multi-spectral scanning module and the dynamic energy control module through a high-speed data interface, including:

[0152] a fractal path generator configured to generate a Hilbert curve and a spiral progressive path based on geometric data of a three-dimensional characteristic spectrum;

[0153] Thermal field simulation unit, with built-in material thermal property database and finite element calculation core, outputs heat accumulation warning signal;

[0154] A motion controller, mechanically connected to the laser scanning galvanometer, receives path instructions and drives the optical actuator;

[0155] Quality feedback correction module, including:

[0156] a confocal microscopy imaging unit optically coupled to the processing area and configured to collect surface topography data of a micro-area;

[0157] The roughness analysis processor is connected to the microscopic imaging unit through an image acquisition card and has a built-in three-dimensional Fourier transform algorithm library;

[0158] The safety margin calculation unit interacts with the dynamic energy control module through a data line and outputs a processing termination instruction;

[0159] The gas-solid separation and recovery module is physically connected to the process chamber outlet and includes:

[0160] High-temperature airflow generating device, integrated with temperature sensor and flow control valve, forms thermodynamic coupling with the thermal field of laser processing area;

[0161] The magnetoelectric composite separation channel has gradient magnetic field generators and electrode arrays arranged in sequence along the direction of material movement;

[0162] Particle monitoring components, including high-speed cameras and image processors, provide feedback to adjust sorting parameters;

[0163] The process optimization subsystem communicates with each module via industrial Ethernet, including:

[0164] A machine learning accelerator card configured to run a deep neural network model and output a process parameter optimization solution;

[0165] Digital twin server, which stores multi-physics field coupling simulation models and compares them with actual processing data in real time;

[0166] Parameter calibration interface, injecting compensation coefficients into the motion controller and laser generating unit;

[0167] The output end of the multi-spectral scanning module transmits the three-dimensional characteristic spectrum to the input end of the dynamic energy regulation module through optical fiber;

[0168] The path planning control module receives the geometric data from the multi-spectral scanning module and the thermodynamic parameters of the dynamic energy control module, and generates a control signal to drive the scanning galvanometer via the motion controller;

[0169] The optical signal input of the quality feedback correction module comes from the reflected light path of the processing area, and the correction instruction output by it is synchronously sent to the dynamic energy regulation module and the path planning control module through the digital I / O interface;

[0170] The airflow parameters of the gas-solid separation and recovery module are controlled by the real-time energy data of the dynamic energy control module, and its particle monitoring signal is fed back to the mass feedback correction module;

[0171] The process optimization subsystem establishes data channels with all modules through the OPC UA protocol to form a closed-loop control network.

[0172] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0173] In this invention, the precise connection of the above steps enables efficient and precise laser lift-off, oxide layer separation, substrate recovery, and surface quality control. In particular, when processing irregularly shaped scrap, the laser energy, scanning path, and thermal control strategy can be dynamically adjusted based on real-time data, effectively avoiding material loss and waste and improving processing efficiency and recovery rates.

[0174] The present invention encompasses any alternatives, modifications, equivalents, and solutions that fall within the spirit and scope of the present invention. While specific details are described in detail in the preferred embodiments to provide a thorough understanding of the present invention, those skilled in the art will be able to fully understand the present invention without these details. Furthermore, to avoid unnecessary confusion regarding the essence of the present invention, well-known methods, processes, procedures, components, and circuits have not been described in detail.

[0175] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A laser positioning stripping method for special-shaped NdFeB scrap, characterized in that: The following steps are involved: S1: Perform multispectral scanning on the surface of special-shaped NdFeB scrap. Through synchronous scanning with long-wavelength laser and short-wavelength laser, the geometric profile data of the matrix material and the thickness distribution data of the oxide layer are obtained respectively. Based on the difference characteristics of the dual-band reflection signal, a three-dimensional characteristic spectrum of the oxide layer-matrix boundary is established; S2: dynamically adjusting the energy parameters of laser stripping based on the oxide layer thickness distribution data in the three-dimensional characteristic spectrum, constructing an energy-thickness matching model, implementing laser stripping using a step-by-step energy control strategy, and monitoring and terminating irradiation in real time through plasma spectroscopy when approaching the substrate interface; S3: generating a fractal scanning path for an adaptive profiled surface based on the geometric profile data of the three-dimensional characteristic spectrum, and dynamically adjusting the scanning path and cooling interval according to real-time thermal accumulation data; S4: Perform in-situ surface roughness testing during the laser lift-off process, trigger the secondary refinement mode based on the test results, and dynamically correct the safety lift-off margin based on the lift-off depth data; S5: The thermal gas flow field generated during the laser stripping process is used in combination with the gas-solid separation dynamics principle to achieve in-situ separation of oxide layer fragments and matrix particles, and the recovery of matrix particles is enhanced by the assistance of a magnetic field.

2. The laser positioning and stripping method for special-shaped NdFeB scrap according to claim 1, characterized in that: Said S1 specifically includes: S1.1: Scan the scrap surface using a long-wavelength laser, where the wavelength of the long-wavelength laser is selected to penetrate the surface oxide layer and reflect the geometric deformation characteristics of the substrate material, to obtain raw point cloud data of the substrate contour; S1.2: Scanning is performed synchronously using a short-wavelength laser, wherein the wavelength of the short-wavelength laser is selected to interact with the electron energy level transition of the oxide layer, and the reflection intensity thereof decays nonlinearly with the thickness of the oxide layer, thereby obtaining optical response data of the oxide layer; S1.3: Establish a phase separation model for the dual-band reflection signals by calculating the difference characteristics between the long-wavelength reflection signal and the short-wavelength reflection signal in the time and frequency domains, including but not limited to the reflection intensity ratio, phase delay difference, and spectral energy distribution difference; S1.4: Construct a dynamic threshold decision mechanism to dynamically adjust the boundary judgment criteria based on the spectral feature distribution of the local area. The judgment criteria include the following steps: The scanning area is divided into several sub-areas, and the characteristic vector of the dual-band reflection signal in each sub-area is calculated; Perform principal component analysis on the eigenvectors to extract the principal component factors that characterize the transition state of the oxide layer matrix; Construct a multidimensional decision space based on principal component factors and use support vector machine algorithm to generate dynamic classification boundaries; S1.5: Convert the discrete optical data into a continuous oxide layer distribution surface using a 3D topology reconstruction engine. The reconstruction process includes: Perform Poisson surface reconstruction on the original point cloud data to generate a matrix geometry model; Mapping the optical response data of the oxide layer to the surface of the geometric model to establish the oxide layer thickness field; The three-dimensional characteristic spectrum of the oxide layer is generated using the non-uniform rational B-spline surface fitting algorithm.

3. The laser positioning and stripping method for special-shaped NdFeB scrap according to claim 1, characterized in that: The S2 specifically includes: S2.1: Construct an energy-thickness matching model, wherein the model satisfies the following conditions: The relationship between the energy parameter and the oxide layer thickness follows an exponential function, and the coefficient of the exponential term is determined by the light-to-heat conversion efficiency of the material; The model includes dual constraints of oxide layer ablation threshold and substrate damage threshold; S2.2: Implement a stepped energy control strategy, including: In the first stage, a high-energy pulse is used to break the dense oxide layer, and the pulse width of the high-energy pulse is selected to meet the pressure threshold required for the impact phase transition; In the second stage, the transition layer is treated in a low-energy, long-pulse mode. The low-energy parameters are set to ensure that the depth of the heat-affected zone is less than the thickness of the interface transition zone between the oxide layer and the substrate. In the third stage, when approaching the matrix interface, the plasma emission spectrum is collected in real time, and the appearance of the matrix material characteristic peak is detected by the characteristic spectrum line recognition algorithm; S2.3: Real-time monitoring of plasma spectrum, including: A fiber optic spectrometer was used to collect the emission spectrum in the wavelength range of 400-700 nm; After wavelet denoising of the spectral data, the relative intensity ratios of the characteristic peaks of rare earth elements were extracted; When the characteristic peak intensity ratio exceeds the preset threshold, the laser output is immediately cut off.

4. The laser positioning and stripping method for special-shaped NdFeB scrap according to claim 1, characterized in that: The S3 specifically includes: S3.1: The fractal scanning path generation method includes: The processing area is decomposed into multi-level substructures, and the filling mode is selected according to the local curvature characteristics. When the curvature radius is less than the critical value, a spiral progressive path is used, otherwise a space filling curve is used; Implementing a recursive subdivision algorithm for high-complexity regions, wherein the subdivision criteria include curvature change rate, oxide thickness gradient, and heat accumulation prediction value; S3.2: Real-time heat accumulation control methods include: Establishing a heat flux density field prediction model, wherein the model calculates the temperature field distribution based on scanning path parameters, material thermal diffusion coefficient and laser energy parameters; Set the heat accumulation warning threshold. When the temperature gradient in a local area exceeds the critical value, the following actions are performed: a: interrupt the current scan path and insert a cooling time interval; b: Switch to the adjacent low-temperature area to continue working; c: Dynamically adjust the filling density of subsequent scanning paths; S3.3: The cooling interval is determined by: Calculate the heat dissipation time according to the heat diffusion equation; Replan the path sequence based on the priority of subsequent processing areas.

5. The laser positioning and stripping method for special-shaped NdFeB scrap according to claim 1, characterized in that: The S4 specifically includes: S4.1: In-situ surface roughness testing methods include: Confocal laser scanning technology is used to perform optical tomography along the normal direction of the peeling area, and micro-area surface morphology data is obtained through axial scanning; Performing a three-dimensional Fourier transform analysis on the tomographic image to extract spatial frequency distribution characteristics, including but not limited to the energy proportion of high-frequency components and the directional intensity distribution of intermediate-frequency components; Establishing a surface roughness evaluation function, wherein the function is calculated by weighting the deviation between the spatial frequency characteristics and a preset reference surface; S4.2: The triggering mechanism of the secondary refinement mode includes: When the surface roughness evaluation function value exceeds the first critical threshold, the finishing mode is started, which includes the following parameter adjustments: a: Reduce the laser power density to a fixed ratio of the original parameters; b: Increase the scanning speed to a fixed multiple of the original speed; c: Reduce the spot overlap rate to the preset safety range; During the finishing process, the surface roughness change rate is continuously monitored and the finishing mode is automatically exited when the change rate approaches zero. S4.3: The method for dynamic correction of the safety stripping margin includes: Establishing a peeling depth prediction model, wherein the model input parameters include cumulative laser energy density, material thermal diffusivity, and oxide layer porosity; The ratio of the remaining substrate thickness to the original thickness is calculated in real time. When the ratio is lower than a second critical threshold, the following operations are performed: a: Immediately terminate the laser irradiation of the current area; b: Mark the area as a prohibited processing area in the three-dimensional characteristic spectrum; c: Replan the scan path density in the surrounding area.

6. The laser positioning and stripping method for special-shaped NdFeB scrap according to claim 1, characterized in that: The specific implementation of S5 includes: S5.1: Gas-solid separation kinetics include: The high-temperature gas generated during the laser ablation process is used to construct a vertical airflow field, and the turbulence intensity of the fluidized bed is controlled by adjusting the gas flow rate. The design of the graded sedimentation chamber is based on the difference in Stokes numbers between the oxide layer fragments and the matrix particles. The difference comes from: The oxide layer fragments have a smaller particle size distribution range; The matrix particles have a higher mass density; A multi-stage baffle is set at the exit of the sedimentation chamber, and the spacing between the baffles is determined according to the terminal velocity distribution of the particles; S5.2: Magnetic field-assisted recovery methods include: A gradient magnetic field generator is arranged at the end of the sorting process, and the magnetic field intensity increases along the direction of particle movement; The matrix particles are pre-magnetized to generate residual magnetic moments through an alternating magnetic field. A particle trajectory control channel is designed based on the Lorentz force principle, wherein the curvature radius of the channel is inversely proportional to the magnetic field gradient; S5.3: Closed-loop control of the sorting process includes: A high-speed camera system is used to monitor the particle movement trajectory in real time; Count the target particle escape rate through image recognition algorithm; Dynamically adjust the air flow speed and magnetic field strength to minimize the escape rate.

7. The laser positioning and stripping method for special-shaped NdFeB scrap according to claim 5, characterized in that: The specific steps of the three-dimensional Fourier transform analysis in S4.1 include: S4.1.1: Convert the tomographic image sequence from the spatial domain to the frequency domain: Perform two-dimensional discrete Fourier transform layer by layer along the XY plane; Perform one-dimensional Fourier transform in the Z-axis direction to generate a three-dimensional spectrum; S4.1.2: Spatial frequency feature extraction method: The three-dimensional spectrum is divided into low-frequency, medium-frequency and high-frequency regions, and the boundary frequencies of each region are determined by the surface characteristics of the material; Calculate the integrated energy of the complex modulus value in each frequency region and normalize it to the proportion of the total energy; Directional filtering is performed on the spectrum in the mid-frequency region, and the energy distribution differences at 0°, 45°, 90°, and 135° are statistically analyzed; S4.1.3: Surface roughness evaluation function construction process: The reference surface is defined as the theoretical frequency distribution curve of the ideal peeling surface; Calculate the root mean square error between the measured data and the reference curve in each frequency band; Different weight coefficients are applied to the errors of each frequency band and then linearly superimposed. The weight coefficients are related to the surface function requirements.

8. The laser positioning and stripping method for special-shaped NdFeB scrap according to claim 6, characterized in that: The design method of the graded settling chamber in S5.1 includes: S5.1.1: Determination of fluidized bed kinetic parameters: Calculate the dynamic viscosity coefficient based on the gas temperature; The formula for the terminal velocity of particles is derived based on Stokes' law, taking into account the influence of particle shape factors; The air flow velocity that maximizes the velocity difference between the oxide layer fragments and the matrix particles is determined by iterative calculation; S5.1.2: Multi-stage baffle optimization method: Establish a differential equation for particle motion trajectory, with input parameters including particle mass, gas flow rate, and baffle geometry parameters; Monte Carlo simulation method is used to calculate the probability distribution of particle collision; Taking the target particle recovery rate as the optimization goal, the optimal solution of the baffle spacing is solved by genetic algorithm; S5.1.3: Turbulence intensity control strategy: A porous medium rectifying layer is provided at the inlet of the settling chamber; Monitor pressure fluctuations in real time and adjust the opening of the air supply valve through the PID controller; The turbulence suppression device is activated when the Reynolds number exceeds a critical value.

9. The laser positioning stripping method for special-shaped NdFeB scrap according to claim 1, characterized in that: The following post-processing steps are also included: S6: Establish a process knowledge base and conduct continuous optimization, including: S6.1: Machine Learning Model Training Methods: Collect historical processing data to build a training set, the data including three-dimensional characteristic spectrum parameters, laser energy sequence, surface roughness test results and sorting efficiency indicators; The defined characteristic vector includes the average value of the oxide layer thickness gradient, the maximum heat accumulation value and the path turning point density; Use deep neural networks to build a multi-objective optimization model and output the optimal process parameter combination; S6.2: Digital Twin Verification Methodology: Build a virtual machining environment and integrate the following physical field simulation modules: a: Thermodynamic model of laser-material interaction; b: Molecular dynamics model of the oxide layer stripping process; c: Computational fluid dynamics model of gas-solid two-phase flow; Input the actual processing parameters into the digital twin system for multi-physics field coupling simulation; When the deviation between the simulation results and the measured data exceeds the allowable range, the model parameter calibration process is triggered; S6.3: Process parameter self-optimization mechanism: Regularly compare and analyze actual production data with the knowledge base; Identify process parameter drift patterns and generate compensation coefficients; The optimization results are injected into the control system through online updating.

10. A laser positioning and stripping system for irregular-shaped NdFeB scrap, used for running the laser positioning and stripping method for irregular-shaped NdFeB scrap according to any of claims 1 to 9, characterized in that ,include: Multispectral scanning module, including: a long wavelength laser scanning unit configured to emit near infrared laser light that penetrates the oxide layer and collects geometric profile data of the substrate; a short-wavelength laser scanning unit configured to emit visible laser light that interacts with electron energy levels of the oxide layer and acquire optical response data of the oxide layer; a signal fusion processing unit, electrically connected to the long-wavelength and short-wavelength laser scanning units, configured to perform phase separation processing on the dual-band reflection signals and generate a three-dimensional characteristic spectrum; The dynamic energy control module is connected to the multi-spectral scanning module via a data bus and includes: an energy calculation engine configured to receive oxide layer thickness data in a three-dimensional characteristic spectrum and calculate a stepped energy parameter; A laser generating unit, electrically connected to the energy calculation engine, comprising an adjustable pulse width fiber laser and a beam shaping component; Plasma monitoring unit, integrating fiber optic spectrometer and characteristic peak recognition circuit, outputs laser termination signal in real time; The path planning control module is connected to the multi-spectral scanning module and the dynamic energy control module through a high-speed data interface; Quality feedback correction module, including: a confocal microscopy imaging unit optically coupled to the processing area and configured to collect surface topography data of a micro-area; The roughness analysis processor is connected to the microscopic imaging unit through an image acquisition card and has a built-in three-dimensional Fourier transform algorithm library; The safety margin calculation unit interacts with the dynamic energy control module through a data line and outputs a processing termination instruction; The gas-solid separation and recovery module is physically connected to the process chamber outlet and includes: High-temperature airflow generating device, integrated with temperature sensor and flow control valve, forms thermodynamic coupling with the thermal field of laser processing area; The magnetoelectric composite separation channel has gradient magnetic field generators and electrode arrays arranged in sequence along the direction of material movement; The particle monitoring component includes a high-speed camera and an image processor, which provides feedback to adjust the sorting parameters.

Images