A galvanometer scanning positioning precision adaptive correction method and system

Abstract

The application discloses a galvanometer scanning positioning precision adaptive correction method and system, relates to the technical field of galvanometer correction, and comprises the following steps: issuing an instruction to a galvanometer system to obtain a measured position signal, comparing the measured signal with a theoretical coordinate to generate an error data set, decoupling according to physical sources into low-order geometric errors and high-order optical distortions, generating a correction parameter set with layered properties and loading correction, generating a positioning deviation amount at a processing gap monitoring reference point, triggering a new error data set after re-calibration and resampling, separating the current parameter set into three components of bias, low order and high order, locking the high-order distortion component inherent to the field mirror, using the bias and low-order components as initial vectors to combine the new error data set for dimension reduction iteration, and generating an updated correction parameter set for replacement. Through the layered decoupling of physical properties and the high-order locking mechanism, the application avoids parameter coupling oscillation, and greatly improves the re-calibration calculation efficiency and convergence speed.

Description

Technical Field

[0001] This invention relates to the field of galvanometer calibration technology, specifically to an adaptive calibration method and system for galvanometer scanning positioning accuracy. Background Technology

[0002] A galvanometer scanning system is a device that changes the direction of beam propagation by controlling the deflection angle of a mirror, thereby enabling the beam to be quickly positioned on a target plane. It typically consists of mutually orthogonal X-axis and Y-axis deflection mirrors, and can deflect the incident beam to any specified coordinate position within the processing area. It is widely used in fields such as laser precision processing, micro-nano manufacturing, and bio-imaging.

[0003] Existing technologies fail to distinguish the physical evolution of system errors, treating low-order linear parameters representing mechanical and electrical offsets and high-order nonlinear distortion parameters representing the inherent characteristics of the field lens as indiscriminate free variables and including them together in the iterative solution space. Since the high-order distortion parameters have a large dimension and high temporal stability within the short period interval between adjacent recalibrations, they are essentially converged static optical models. Performing indiscriminate global iterative optimization on such converged stable parameters and easily drifting low-order parameters not only leads to a sharp expansion of the solution space and consumes a lot of computational resources, making it difficult to complete the recalibration operation in real time within the strict processing intervals, but also causes unnecessary perturbations of high-order parameters to destroy their established distortion compensation benchmarks, causing coupled oscillations between low-order and high-order parameters. This makes it difficult for the updated correction parameters to converge to the optimal solution quickly, ultimately causing the failure of the online dynamic adaptive correction mechanism. Summary of the Invention

[0004] To address the aforementioned technical problems, this paper provides an adaptive correction method and system for galvanometer scanning positioning accuracy. This solution overcomes the shortcomings of existing technologies, which fail to distinguish the physical evolution of system errors. They treat low-order linear parameters representing mechanical and electrical offsets and high-order nonlinear distortion parameters representing the inherent characteristics of the field lens as indiscriminate free variables, incorporating them into the iterative solution space. Since high-order distortion parameters are dimensionally large and possess high temporal stability within short intervals between adjacent recalibrations, they essentially belong to a converged static optical model. Undifferentiated global iterative optimization of such converged stable parameters and easily drifting low-order parameters not only leads to a rapid expansion of the solution space and consumes significant computational resources, making recalibration difficult to complete in real-time within stringent processing intervals, but also causes unnecessary perturbations of high-order parameters to disrupt their established distortion compensation benchmarks, triggering coupled oscillations between low-order and high-order parameters. This results in the updated correction parameters failing to converge quickly to the optimal solution, ultimately causing the failure of the online dynamic adaptive correction mechanism.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: An adaptive correction method for galvanometer scanning positioning accuracy, specifically including: S1. Send control commands corresponding to the theoretical coordinates to the galvanometer system, rely on the detection beam to be deflected by the deflection mirror, extract part of the beam from the deflected optical path through the beam splitter and guide it to the position detection device, and obtain the measured position signal output by the position detection device corresponding to the physical position of the laser spot. S2. Compare the measured position signal with the theoretical coordinates to obtain a coordinate error dataset, input the coordinate error dataset into the compensation algorithm, calculate and output a set of correction parameters for correcting the control command; S3. Load the correction parameter set into the galvanometer control link, correct the deflection angle of the subsequent control commands, and during the processing after the correction parameter set is loaded, control the probe beam to be projected onto the preset reference sampling point located in the non-processing area at the edge of the processing area to generate the positioning deviation. S4. When the positioning deviation exceeds the threshold, control the probe beam to traverse the regular dot matrix sequence within the processing area and re-traverse the sampling points to generate a coordinate error dataset. Separate the currently loaded correction parameter set into bias components, low-order distortion components and high-order distortion components. S5. Extract the bias component and low-order distortion component as initial parameters for iteration, and keep the high-order distortion component unchanged. Perform iterative solution based on the initial parameters for iteration and the regenerated coordinate error dataset to generate an updated correction parameter set and perform loading.

[0006] Preferably, step S1 specifically includes: A reference coordinate system is established based on the working area of ​​the galvanometer scanning, and a regular point matrix sequence covering the working area is delineated within the reference coordinate system. The coordinate values ​​of each theoretical coordinate point in the regular point matrix sequence are then obtained. According to the order of the rule lattice sequence, control commands corresponding to each theoretical coordinate point are sequentially sent to the galvanometer system to control the probe beam to be deflected to the corresponding theoretical position by the deflection mirror; Extract a portion of the beam from the deflected optical path and guide it to the position detection device; After performing noise reduction processing by averaging multiple samples on the electrical signal output by the position detection device, the centroid algorithm is used to calculate the measured coordinates of the physical position of the laser spot, which are then used as the measured position signal.

[0007] Preferably, step S2 specifically includes: The set of coordinate deviations corresponding to the measured position signals of all sampling points is called the coordinate error dataset. The coordinate error dataset is decoupled into low-order geometric error components and high-order optical distortion components according to the physical source of the error. For the low-order geometric error components, an affine transformation matrix containing two-axis scaling factors and orthogonal bias parameters is constructed. The affine transformation matrix is ​​solved by the least squares method to generate low-order correction parameters for compensating for mechanical and electrical offsets. For the higher-order optical distortion components, a distortion grid table covering the scanning area is constructed for the residual spatial deviation after deducting the lower-order geometric error components. Spatial interpolation is performed on the distortion grid table to generate higher-order correction parameters for compensating for the inherent distortion of the field lens. The low-order correction parameters and the high-order correction parameters are merged according to a preset data structure to output a set of correction parameters with hierarchical attributes.

[0008] Preferably, step S3 specifically includes: The set of correction parameters is loaded into the real-time command generation stage of the galvanometer controller. Each time a new control command is generated, the current theoretical coordinates are superimposed with the compensation values ​​corresponding to the set of correction parameters to obtain the corrected actual execution coordinates. After the set of correction parameters is successfully loaded and enters the processing process, during the idle period when the current processing trajectory is completed and the next processing trajectory has not yet started, the probe beam is controlled to be projected at a probe power that does not produce a substantial processing effect onto at least one preset reference sampling point located at the edge of the set processing area and outside the area occupied by the current actual processing graphic. The measured position signal output by the position detection device at the preset reference sampling point is obtained, and the difference between the two-dimensional coordinates corresponding to the measured position and the theoretical two-dimensional coordinates of the preset reference sampling point is calculated to obtain the two-dimensional positioning deviation, which is used as the positioning deviation representing the current positioning state of the galvanometer system.

[0009] Preferably, step S4 specifically includes: When the positioning deviation exceeds the preset threshold, i.e. the recalibration trigger condition is met, the probe beam is controlled to be projected sequentially onto each sampling point in the processing area according to the preset sampling point set. The sampling point set is completely consistent with the regular dot matrix sequence used when the correction parameter set was initially generated, or a simplified dot matrix sequence formed by selecting intervals of the regular dot matrix sequence to reduce data acquisition time. The measured position signal corresponding to each sampling point is obtained by the position detection device, and the difference between the physical coordinates corresponding to the measured position signal and the theoretical coordinates is compared to regenerate the coordinate error dataset in the current state. The currently loaded set of correction parameters is structured and analyzed based on the physical error sources and spatial variation characteristics corresponding to each parameter item in the set of correction parameters. It is divided into offset components that represent the overall translation of the coordinates, low-order distortion components that represent the scaling and angular deviation of the two axes, and high-order distortion components that represent the inherent distortion of the optical system due to non-uniform spatial distribution.

[0010] Preferably, step S5 specifically includes: The bias component and the low-order distortion component are extracted as the initial parameter vector for iterative solution, wherein the bias component provides an initial estimate of the current system zero-point translation state, and the low-order distortion component provides an initial estimate of the current system two-axis scaling factor and orthogonal deflection state. The higher-order distortion components are kept locked during this iterative solution process and are regarded as converged intrinsic optical distortion parameters, and are not included in the parameter set to be optimized. Using the regenerated coordinate error dataset as the observations in the objective function, and taking the initial parameter vector as the search starting point, a preset iterative optimization algorithm is executed to solve for the update bias component and update low-order distortion component that minimize the residual. The updated bias component, the updated low-order distortion component, and the unchanged high-order distortion component are merged to generate the updated correction parameter set, and the updated correction parameter set is written into the galvanometer control link to replace the currently loaded correction parameter set.

[0011] An adaptive correction system for galvanometer scanning positioning accuracy, used to implement the above-mentioned adaptive correction method, specifically includes: The optical path guidance and calculation module is used to guide the deflection of the probe beam in response to theoretical coordinate control commands. After beam splitting and extraction, the beam spot signal is obtained through the position detection device, and the signal is denoised and centroid calculated to output the measured position signal. The error decoupling calibration module is used to compare the measured position signal with the theoretical coordinates to generate an error dataset, and then decouple it into low-order geometric error and high-order optical distortion according to the physical source. The low-order and high-order correction parameters are generated by matrix solving and spatial interpolation respectively, and the correction parameter set is merged and output. The instruction compensation monitoring module is used to superimpose the set of correction parameters into the real-time control instruction to correct the deflection angle, and control the probe beam to be projected onto the reference point in the non-processing area during the processing gap, and extract the positioning deviation representing the current system drift state by comparison. The adaptive recalibration deduction module is used to respond to the positioning deviation triggering the resampling of a new error dataset, and to structurally separate the current correction parameter set into three components: bias, low-order and high-order. The high-order component is locked and iteratively solved using only the bias and low-order components as initial vectors in combination with the new error dataset. This generates an updated correction parameter set to replace the current parameters in order to achieve error closed-loop convergence.

[0012] Preferably, the optical path guidance and resolution module includes: The coordinate establishment unit is used to establish a reference coordinate system based on the working area of ​​the galvanometer scanning, and to delineate a regular dot matrix sequence covering the working area within the reference coordinate system to obtain the coordinate values ​​of each theoretical coordinate point. The deflection control unit is used to send control commands corresponding to each theoretical coordinate point to the galvanometer system in the order of the rule dot matrix sequence, so as to control the probe beam to be deflected to the corresponding theoretical position by the deflection mirror. The beam splitting and extraction unit is used to extract a portion of the beam from the deflected optical path and guide it to the position detection device. The noise reduction calculation unit is used to perform noise reduction processing on the electrical signal output by the position detection device by multiple sampling and averaging, and then use the centroid algorithm to calculate the measured coordinates of the physical position of the laser spot, which is then output as the measured position signal.

[0013] Preferably, the error decoupling calibration module includes: The error comparison unit is used to collect the coordinate deviations corresponding to the measured position signals of all sampling points into a coordinate error dataset. The physical decoupling unit is used to decouple the coordinate error dataset into low-order geometric error components and high-order optical distortion components according to the physical source of the error. The low-order solving unit is used to construct an affine transformation matrix containing a two-axis scaling factor and an orthogonal bias parameter for the low-order geometric error components, and solve the affine transformation matrix by the least squares method to generate low-order correction parameters for compensating for mechanical and electrical offsets. The higher-order interpolation unit is used to construct a distortion grid table covering the scanning area for the residual spatial deviation after deducting the lower-order geometric error components, and to perform spatial interpolation operations on the distortion grid table to generate higher-order correction parameters for compensating for the inherent distortion of the field lens. The parameter merging unit is used to merge the low-order correction parameters and the high-order correction parameters according to a preset hierarchical data structure, and output a set of correction parameters with hierarchical attributes.

[0014] Preferably, the instruction compensation monitoring module includes: The instruction superposition unit is used to load the correction parameter set into the real-time instruction generation stage of the galvanometer controller. Each time a new control instruction is generated, the current theoretical coordinates are superimposed with the compensation values ​​corresponding to the correction parameter set to obtain the corrected actual execution coordinates. The reference projection unit is used to control the probe beam to project at a probe power that does not produce a substantial processing effect onto a preset reference sampling point located in the non-processing area at the edge of the processing area during the idle period when the current processing trajectory has been completed and the next processing trajectory has not yet started. The deviation extraction unit is used to acquire the measured position signal output by the position detection device at the preset reference sampling point, calculate the difference between the measured position and the theoretical coordinates, and obtain the positioning deviation amount representing the current system drift state. The adaptive recalibration inference module includes: The resampling trigger unit is used to respond to the positioning deviation exceeding the threshold, control the detection beam to retrace the sampling points according to the preset sampling point set, obtain the measured position signal of each sampling point through the position detection device, compare the difference between the measured physical coordinates and the theoretical coordinates, and regenerate the coordinate error dataset. The structure stripping unit is used to perform structured analysis on the currently loaded set of correction parameters, dividing it into offset components representing the overall coordinate translation, low-order distortion components representing the scaling and angular deviation of the two axes, and high-order distortion components representing the inherent distortion of the optical system due to non-uniform spatial distribution. The locking iteration unit is used to keep the higher-order distortion components locked during the iterative solution process, treating them as converged intrinsic optical distortion parameters. The offset component and the lower-order distortion components are used as the initial vectors for iteration, and the dimensionality reduction iterative optimization is performed in combination with the newly generated coordinate error dataset to solve for the updated offset component and the updated lower-order distortion components. The parameter merging and overwriting unit is used to merge the updated bias component, the updated low-order distortion component, and the unchanged high-order distortion component into an updated correction parameter set, and write it into the galvanometer control link to replace the current parameters, thereby achieving error closed-loop convergence.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: In the actual processing after the calibration parameters are loaded, this invention continuously acquires the measured position signal output by the position detection device to monitor the positioning deviation of the galvanometer. When the positioning deviation meets the preset recalibration triggering conditions, the generation and loading steps of the calibration parameter set are automatically re-executed. The quantitative diagnosis of the system drift can be completed within the processing interval without interrupting normal production. This provides an objective and quantifiable decision basis for recalibration triggering, avoiding the blindness of relying on manual experience judgment or triggering recalibration at fixed time periods in the prior art.

[0016] When re-executing the calibration parameter set generation step, this invention uses the currently loaded calibration parameter set as the initial parameters for the compensation algorithm to iteratively solve the problem. This hot-start mechanism ensures that the initial parameters are already in the neighborhood of the new optimal solution. The iterative algorithm only needs a few steps to reconverge to the new optimal solution. While ensuring accuracy, it significantly reduces the computation time of recalibration and solves the problem of discarding all historical calibration information and having to start from scratch for each recalibration in the prior art. This allows the recalibration operation to be completed quickly within the processing interval.

[0017] In the initial calibration stage of the correction parameters, this invention decouples the system error into low-order geometric error components and high-order optical distortion components according to their physical sources. These components are then processed using the least-squares solution of the affine transformation matrix and spatial interpolation of the distortion grid table, respectively. The results are then encapsulated into a data structure with hierarchical attributes, ensuring that each parameter corresponds one-to-one with its corresponding physical error source. This provides direct data structure support for quickly separating the components affected by temperature drift and the stable components during recalibration. Attached Figure Description

[0018] Figure 1 This is a flowchart of the adaptive correction method in this invention; Figure 2 This is a flowchart of error decoupling in the adaptive correction method of this invention; Figure 3 This is a flowchart of the adaptive correction process in the adaptive correction method of this invention; Figure 4 This is a system framework diagram of the adaptive correction system in this invention. Detailed Implementation

[0019] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0020] Reference Figures 1-3 As shown, an adaptive correction method for galvanometer scanning positioning accuracy specifically includes: S1. Send control commands corresponding to the theoretical coordinates to the galvanometer system, rely on the detection beam to be deflected by the deflection mirror, extract part of the beam from the deflected optical path through the beam splitter and guide it to the position detection device, and obtain the measured position signal output by the position detection device corresponding to the physical position of the laser spot. S2. Compare the measured position signal with the theoretical coordinates to obtain a coordinate error dataset, input the coordinate error dataset into the compensation algorithm, calculate and output a set of correction parameters for correcting the control command; S3. Load the correction parameter set into the galvanometer control link, correct the deflection angle of the subsequent control commands, and during the processing after the correction parameter set is loaded, control the probe beam to be projected onto the preset reference sampling point located in the non-processing area at the edge of the processing area to generate the positioning deviation. S4. When the positioning deviation exceeds the threshold, control the probe beam to traverse the regular dot matrix sequence within the processing area and re-traverse the sampling points to generate a coordinate error dataset. Separate the currently loaded correction parameter set into bias components, low-order distortion components and high-order distortion components. S5. Extract the bias component and low-order distortion component as initial parameters for iteration, and keep the high-order distortion component unchanged. Perform iterative solution based on the initial parameters for iteration and the regenerated coordinate error dataset to generate an updated correction parameter set and perform loading.

[0021] Preferably, step S1 specifically includes: A reference coordinate system is established based on the working area of ​​the galvanometer scanning, and a regular point matrix sequence covering the working area is delineated within the reference coordinate system. The coordinate values ​​of each theoretical coordinate point in the regular point matrix sequence are then obtained. According to the order of the rule lattice sequence, control commands corresponding to each theoretical coordinate point are sequentially sent to the galvanometer system to control the probe beam to be deflected to the corresponding theoretical position by the deflection mirror; Extract a portion of the beam from the deflected optical path and guide it to the position detection device; After performing noise reduction processing by averaging multiple samples on the electrical signal output by the position detection device, the centroid algorithm is used to calculate the measured coordinates of the physical position of the laser spot, which are then used as the measured position signal. In this scheme, based on the preset working area range of the galvanometer scanning system, a two-dimensional Cartesian coordinate system is established with the geometric center of the working area as the origin of the coordinate system, the X-axis deflection direction of the galvanometer as the horizontal axis and the Y-axis deflection direction as the vertical axis. This system serves as the unified reference coordinate system for all subsequent coordinate calculations and error measurements. Within the reference coordinate system, a regular dot matrix sequence covering the working area is defined, wherein the regular dot matrix sequence adopts an equally spaced rectangular grid dot matrix, and the spacing between adjacent sampling points is determined according to the size of the working area and the required calibration resolution. The theoretical coordinate values ​​of each sampling point within the reference coordinate system are calculated and stored. The spacing between adjacent sampling points is determined based on the working area size and the required calibration resolution. For example, for a working area of ​​400mm×400mm, the sampling point spacing is set to 20mm, forming a regular dot matrix of 21×21 sampling points with a total of 441 sampling points. The higher the calibration resolution requirement, the smaller the spacing. The selection of the spacing also needs to meet the requirement that the number of sampling points should be no less than twice the minimum number of sampling points required to solve the low-order affine transformation matrix, so as to ensure the numerical stability of affine fitting in subsequent error decoupling. According to the order of the rule dot matrix sequence, control commands corresponding to each theoretical coordinate point are sequentially sent to the galvanometer system. The control commands include X-axis deflection angle values ​​and Y-axis deflection angle values. The angle values ​​are generated by converting the theoretical coordinate values ​​through the standard coordinate-angle mapping relationship of the galvanometer control system. After receiving each set of control commands, the probe beam is reflected sequentially by the X-axis deflection mirror and the Y-axis deflection mirror, and deflected to the spatial angle position corresponding to the theoretical coordinates. From the optical path after deflection by the deflector, a portion of the beam is extracted using a beam splitter located in the optical path between the deflector and the field mirror. The beam splitting ratio is set based on the sensitivity of the position detection device and the total power of the detection beam: the power of the beam extracted to the position detection device should be greater than the noise equivalent power threshold of the position detection device, and the power of the beam transmitted to the subsequent optical path should be maintained above 95% of the total power of the detection beam. The 95% transmission lower limit is set based on the following: when the noise equivalent power of the position detection device is not higher than 1% of the total power of the detection beam, extracting no more than 5% of the beam power is sufficient to meet the signal-to-noise ratio requirements. A portion of the extracted light beam is guided to a position detection device. The photosensitive surface of the position detection device is in an optically conjugate position with the focal plane of the field lens, so that the lateral position of the measurement spot focused on the photosensitive surface has a definite proportional mapping relationship with the lateral position of the corresponding laser spot on the processing plane. The tolerance of the conjugate relationship is determined according to the system accuracy requirements: when the system requires the positioning error correction accuracy to reach the micrometer level, the axial tolerance of the photosensitive surface deviating from the conjugate focal plane should be controlled within 1 / 4 of the Rayleigh length to avoid the position shift of the measurement spot due to defocusing. The electrical signal output by the position detection device is subjected to noise reduction processing by averaging multiple samples, and the number of samples is limited to 256. After noise reduction, the centroid algorithm is used to calculate the measured coordinates of the physical position of the laser spot. The centroid algorithm uses the signal intensity of each pixel on the photosensitive surface of the position detection device as the weight to calculate the first moment of the energy distribution of the spot, which is used as the measured physical position coordinates of the spot. The measured coordinates are based on the reference coordinate system and are used as the measured position signal output. When a portion of the light spot exceeds the effective detection range of the photosensitive surface of the position detection device, the measured signal of the sampling point is considered invalid. The validity threshold is determined based on the relationship between the light spot diameter and the size of the photosensitive surface: when the distance from the center coordinate of the light spot to the edge of the photosensitive surface is less than 1.2 times the radius of the light spot, it is determined that the light spot may have been truncated, and the corresponding sampling point data is not included in the subsequent error dataset.

[0022] Preferably, step S2 specifically includes: The set of coordinate deviations corresponding to the measured position signals of all sampling points is called the coordinate error dataset. The coordinate error dataset is decoupled into low-order geometric error components and high-order optical distortion components according to the physical source of the error. For the low-order geometric error components, an affine transformation matrix containing two-axis scaling factors and orthogonal bias parameters is constructed. The affine transformation matrix is ​​solved by the least squares method to generate low-order correction parameters for compensating for mechanical and electrical offsets. For the higher-order optical distortion components, a distortion grid table covering the scanning area is constructed for the residual spatial deviation after deducting the lower-order geometric error components. Spatial interpolation is performed on the distortion grid table to generate higher-order correction parameters for compensating for the inherent distortion of the field lens. The low-order correction parameters and the high-order correction parameters are merged according to a preset data structure to output a set of correction parameters with hierarchical attributes; In this scheme, the measured signal collected by the position detection device is compared with the coordinates of the theoretical coordinate points to obtain the two-dimensional position deviation vector of each sampling point, and then merged into a coordinate error dataset. Subsequently, the error distribution is solved by the least squares affine fitting and affine transformation model, which is the low-order geometric error component. The original coordinate deviation of each sampling point is subtracted one by one from the fitted value of the low-order geometric error component at the corresponding sampling point to obtain the residual spatial deviation of each sampling point. The process of solving the error distribution using the least squares affine fitting and affine transformation model is as follows: Let the theoretical coordinates of the sampling point be (X, Y), and the corresponding two-dimensional position deviation vector at the sampling point be (ΔX, ΔY). The affine model expresses the deviation as a linear function of the theoretical coordinates: ; In the formula, the parameter Represents the scaling factor of the two axes and the orthogonal deflection between the axes, parameters The rigid translation amount representing the origin of the coordinate system is fully described by the six-parameter model, which comprehensively describes the zero-order translation and first-order linear transformation errors that may exist in the galvanometer system. Substituting the theoretical coordinates and corresponding deviation vectors of all N regular point lattice sequences, an overdetermined linear equation system containing 2N equations and 6 unknowns is constructed. The overdetermined equation system is solved using the least squares method to obtain the optimal parameter estimate that minimizes the sum of squared residuals of all sampling points. The error distribution described by the affine transformation model obtained by the least squares solution is the low-order geometric error component. The residual spatial deviation of all sampling points is the higher-order optical distortion component. The higher-order optical distortion component physically corresponds to the nonlinear distortion introduced by the field lens due to optical design and manufacturing tolerances. It mainly includes pincushion distortion and barrel distortion. For the decoupled lower-order geometric error component, based on the solved affine model parameters, an affine transformation matrix and offset vector are constructed. Through matrix inversion, lower-order correction parameters are generated to compensate for the lower-order geometric error component. During online compensation, for any theoretical coordinate, the lower-order correction parameters are first applied for linear correction to obtain intermediate corrected coordinates, which are then processed by the higher-order correction stage. For the decoupled higher-order optical distortion components, a distortion grid table covering the scanning area is constructed based on the residual spatial deviation data. The construction method is as follows: the coordinate position of each sampling point is used as the grid node, and the residual spatial deviation at the node is used as the storage value of the node to form a discrete two-dimensional distortion mapping table. The grid spacing of the distortion grid table is the spacing of the sampling point array, and the grid coverage area is the working area covered by the regular point array sequence. For invalid grid nodes that lack measured data due to spot truncation or other reasons, the node storage value is set to zero or filled with the linear extrapolation value of the adjacent valid node to ensure the matrix integrity of the distortion grid table. Spatial interpolation is performed on the distorted mesh table to generate higher-order correction parameters at any intermediate position. The spatial interpolation operation uses bilinear interpolation or bicubic spline interpolation. The selection of the interpolation algorithm is determined based on the measurement noise level of the position detection device and the required spot positioning accuracy: when the residual noise amplitude after the noise reduction processing of the single measurement noise of the position detection device is greater than 1 / 3 of the system accuracy index, bilinear interpolation can be used; when the residual noise amplitude is less than 1 / 3 of the accuracy index, bicubic spline interpolation is used to ensure the curvature continuity at the interpolation nodes and to more accurately approximate the smooth physical characteristics of the field mirror distortion. The generated low-order correction parameters and high-order correction parameters are merged according to a preset hierarchical data structure. The hierarchical data structure clearly distinguishes two parameter levels: the first level is the low-order correction parameter level, which stores the six independent parameters of the affine transformation matrix; the second level is the high-order correction parameter level, which stores the grid node coordinates of the distortion grid table and the two-dimensional residual deviation values ​​of each node, as well as the type identifier of the interpolation algorithm used, to obtain the correction parameter set after merging.

[0023] Preferably, step S3 specifically includes: The set of correction parameters is loaded into the real-time command generation stage of the galvanometer controller. Each time a new control command is generated, the current theoretical coordinates are superimposed with the compensation values ​​corresponding to the set of correction parameters to obtain the corrected actual execution coordinates. After the set of correction parameters is successfully loaded and enters the processing process, during the idle period when the current processing trajectory is completed and the next processing trajectory has not yet started, the probe beam is controlled to be projected at a probe power that does not produce a substantial processing effect onto at least one preset reference sampling point located at the edge of the set processing area and outside the area occupied by the current actual processing graphic. The measured position signal output by the position detection device at the preset reference sampling point is obtained, and the difference between the two-dimensional coordinates corresponding to the measured position and the theoretical two-dimensional coordinates of the preset reference sampling point is calculated to obtain the two-dimensional positioning deviation, which is used as the positioning deviation representing the current positioning state of the galvanometer system. In this scheme, the generated set of correction parameters is loaded into the real-time command generation stage of the galvanometer controller. The set of correction parameters includes a low-order correction parameter layer and a high-order correction parameter layer. The low-order correction parameter layer stores the inverse affine transformation matrix and the inverse offset vector. The high-order correction parameter layer stores the coordinates of each node in the distorted mesh table, the two-dimensional residual deviation value of each node, and the interpolation algorithm type identifier. After the calibration parameter set is successfully loaded and enters the normal processing process, online monitoring is performed during the idle period of the processing trajectory. The criteria for determining the idle period are: all scanning segments of the current processing trajectory have been completed, the laser output has been switched to the off state or the standby state with the maintenance power lower than the processing threshold, and the next processing trajectory instruction sequence in the galvanometer control card instruction buffer has not yet started to be executed. Each time a new control command is generated, the theoretical coordinates of the current path target point are read. First, the inverse affine transformation matrix and the reverse offset vector are extracted from the low-order correction parameter layer, and the theoretical coordinates are linearly corrected to obtain intermediate corrected coordinates. Then, from the high-order correction parameter layer, based on the position of the intermediate corrected coordinates in the distortion grid table, the residual deviation value corresponding to that position is extracted through spatial interpolation. The intermediate corrected coordinates are superimposed with the corresponding residual deviation value to obtain the final actual execution coordinates. The actual execution coordinates are converted into X-axis deflection angle values ​​and Y-axis deflection angle values ​​according to the coordinate-angle mapping relationship preset by the galvanometer system, and are used as the corrected control command output. The preset coordinate-angle mapping relationship is determined by the deflection sensitivity of the galvanometer motor and the optical path geometry. After the calibration parameter set is successfully loaded and the normal processing begins, the online monitoring action is initiated during the idle period when the current processing trajectory has been completed and the next processing trajectory has not yet started. During the idle period, the detection beam is controlled to be projected onto a preset reference sampling point at the detection power, ensuring that the power that generates only the spot signal required for position detection without causing substantial processing effects on the workpiece is the detection power; The preset reference sampling points are located in the edge area of ​​the processing area and outside the area occupied by the current actual processing graphic. The edge area of ​​the processing area refers to the annular area inward from the geometric boundary of the processing area. The horizontal and vertical coordinates of the reference sampling points are pre-selected so that they do not fall into the projection area of ​​the current processing task's processing graphic within the processing area. The number of reference sampling points is at least one. When multiple reference sampling points are set, each reference sampling point is symmetrically distributed at the four corners or the midpoints of the four sides of the processing area. The detection beam is controlled to be projected sequentially to each preset reference sampling point, and the electrical signal output by the position detection device at each reference point is obtained. The same multiple sampling, averaging, noise reduction processing and centroid algorithm calculation as in step S1 are performed to obtain the measured two-dimensional coordinates at each reference point. The difference between the measured two-dimensional coordinates and the theoretical two-dimensional coordinates of the reference point is calculated as the two-dimensional positioning deviation vector. When multiple reference points are set, the maximum value of the magnitude or root mean square value of the two-dimensional deviation vector of each reference point is used as the positioning deviation amount representing the current positioning state of the galvanometer system. The positioning deviation amount is compared with a preset drift safety threshold. The drift safety threshold is set based on the measurement repeatability accuracy limit of the position detection device and the maximum position tolerance zone allowed by the current processing task. When the positioning deviation amount exceeds the drift safety threshold, it is determined that the current positioning state of the galvanometer system has deteriorated, and a recalibration trigger signal is output.

[0024] Preferably, step S4 specifically includes: When the positioning deviation exceeds the preset threshold, i.e. the recalibration trigger condition is met, the probe beam is controlled to be projected sequentially onto each sampling point in the processing area according to the preset sampling point set. The sampling point set is completely consistent with the regular dot matrix sequence used when the correction parameter set was initially generated, or a simplified dot matrix sequence formed by selecting intervals of the regular dot matrix sequence to reduce data acquisition time. The measured position signal corresponding to each sampling point is obtained by the position detection device, and the difference between the physical coordinates corresponding to the measured position signal and the theoretical coordinates is compared to regenerate the coordinate error dataset in the current state. The currently loaded set of correction parameters is structured and analyzed based on the physical error sources and spatial variation characteristics corresponding to each parameter item in the set of correction parameters. It is divided into offset components that represent the overall translation of the coordinates, low-order distortion components that represent the scaling and angular deviation of the two axes, and high-order distortion components that represent the inherent distortion of the optical system that is not uniformly distributed in space. In this scheme, when the positioning deviation exceeds the drift safety threshold, it is determined that the recalibration trigger condition is met, and the recalibration process is started. After the recalibration process is started, the probe beam is controlled to be projected onto the regular dot matrix sequence within the processing area according to the preset sampling point set. The sampling point set provides two optional modes: the first mode is completely consistent with the regular dot matrix sequence used when the correction parameter set is initially generated. The basis is to retain the complete spatial sampling frequency to ensure sufficient resolution capability for any possible order error. It is suitable for scenarios with extremely high accuracy requirements and allowance for long downtime. The second mode is a simplified dot matrix sequence formed by selecting intervals from the regular dot matrix sequence. The basis is that the thermal drift of the galvanometer system is mainly manifested as low-order linear shift and rotation. The changes of such low-frequency spatial errors within the area are extremely gradual, and they can be accurately fitted without high-frequency sampling points. Therefore, by reasonably reducing the number of sampling points, the data acquisition time is greatly shortened, and the impact of recalibration on the processing cycle is reduced. The choice between the two modes is automatically determined based on the relationship between the current cumulative processing time and the allowable time for a single recalibration. The measured position signal corresponding to each sampling point is obtained by the position detection device. The difference between the calculated measured physical coordinates and the corresponding theoretical coordinates is compared to regenerate the coordinate error dataset under the current thermal drift state. The coordinate error dataset reflects the comprehensive deviation between the current actual positioning position and the theoretical position of the galvanometer system after thermal drift. Finally, the structured parsing and component partitioning steps of the loaded calibration parameter set are performed. Since the calibration parameter set generated in the initial calibration is stored in a data structure with hierarchical attributes, the current loaded calibration parameter set can be directly reversed and structured parsed to read the bias components, low-order distortion components, and high-order distortion components. Among the six independent parameters stored in the low-order calibration parameter layer, two parameters representing the X and Y axis translations are extracted as bias components, and four matrix elements representing the ratio and orthogonal deflection of the two axes are extracted as low-order distortion components. The data of the high-order calibration parameter layer is directly used as high-order distortion components.

[0025] Preferably, step S5 specifically includes: The bias component and the low-order distortion component are extracted as the initial parameter vector for iterative solution, wherein the bias component provides an initial estimate of the current system zero-point translation state, and the low-order distortion component provides an initial estimate of the current system two-axis scaling factor and orthogonal deflection state. The higher-order distortion components are kept locked during this iterative solution process and are regarded as converged intrinsic optical distortion parameters, and are not included in the parameter set to be optimized. Using the regenerated coordinate error dataset as the observations in the objective function, and taking the initial parameter vector as the search starting point, a preset iterative optimization algorithm is executed to solve for the update bias component and update low-order distortion component that minimize the residual. The updated bias component, the updated low-order distortion component, and the unchanged high-order distortion component are merged to generate the updated correction parameter set, and the updated correction parameter set is written into the galvanometer control link to replace the currently loaded correction parameter set. In this scheme, the bias component and the low-order distortion component are extracted as the initial parameter vector for iterative solution. Two translation parameters are read from the bias component, and four elements of the inverse affine transformation matrix are read from the low-order distortion component. The above six parameters are arranged into a six-dimensional initial parameter vector according to the encoding order of translation parameters first and matrix elements last. The initial parameter vector provides the search starting point for the subsequent iterative optimization algorithm. The translation parameters in the bias component provide the initial estimate of the current zero-point translation state of the system, and the matrix elements in the low-order distortion component provide the initial estimate of the current system's two-axis scaling factor and inter-axis orthogonal deflection state. The higher-order distortion components are locked during the iterative solution process. In the set of parameters to be optimized in the iterative optimization algorithm, only the bias parameters and low-order distortion parameters contained in the six-dimensional initial parameter vector are included. The distortion grid table node data contained in the higher-order distortion components are not included in the list of variables to be optimized. During the entire iterative solution process, the residual values ​​of each node of the higher-order distortion components remain completely consistent with those extracted analytically and do not participate in any iterative update calculations. The higher-order distortion components characterize the inherent optical distortion characteristics of the field lens, which are determined by the optical design and manufacturing tolerances of the field lens. They have high temporal stability within the time scale covered by recalibration and can be regarded as converged static optical model parameters. Each sampling point in the regenerated coordinate error dataset is taken as the observation value. The residual between the predicted value and the observed value of the error compensation model described by the six-dimensional parameter vector to be optimized at each sampling point is taken as the optimization objective. The objective function is defined as the sum of squares of the two-dimensional residual vectors of all sampling points. The optimization objective is to find the six-dimensional parameter vector that minimizes the objective function value. Starting with the initial parameter vector, a preset iterative optimization algorithm is executed. The iterative optimization algorithm adopts the Gauss-Newton method. In each step of the iteration process, the iterative optimization algorithm calculates the gradient or approximate second derivative information of the objective function based on the current parameter vector, determines the update direction and step size of the parameter vector, and generates the next set of parameter vectors. Since the initial parameter vector comes from the currently loaded calibration parameter set, and the current calibration parameter set is the parameter that has converged to the optimal solution in the previous calibration or recalibration, and the system state change between two adjacent recalibrations is usually small, the initial parameter vector is already in the neighborhood of the optimal solution of the objective function. The iterative algorithm only needs a few steps to reconverge to the new optimal solution. After each iteration of updating the parameter vector, the change in the current objective function value relative to the previous iteration is calculated. When the relative decrease in the objective function value is lower than the preset convergence threshold, or when the number of iterations reaches the preset maximum number of iterations of 10, the iteration is terminated and the current parameter vector is output as the solution result. The updated bias component and the updated low-order distortion component are extracted from the converged six-dimensional parameter vector. The first two dimensions are extracted as the updated bias component, and the last four dimensions are extracted as the updated low-order distortion component. The updated bias component, the updated low-order distortion component, and the high-order distortion component locked in the second step are merged. The merging method adopts the same hierarchical data structure as step S2: the first layer stores the updated bias component and the updated low-order distortion component, and the second layer stores the untouched high-order distortion component. The updated correction parameter set generated after merging is completely consistent with the initially generated correction parameter set in terms of data structure, and the hierarchical attributes remain unchanged. The updated calibration parameter set is written to the calibration parameter storage area of ​​the galvanometer control link, overwriting the currently loaded old calibration parameter set. After the writing is completed, the real-time command generation stage of the galvanometer controller will automatically use the new calibration parameters to correct the deflection angle in the generation of subsequent control commands, thereby completing a complete adaptive recalibration closed loop. The convergence threshold is set based on the fact that when the decrease in the objective function value is less than the minimum objective function fluctuation that can be caused by measurement noise, there is no practical significance to continue iterating, and at this point it is determined that convergence has been achieved. The maximum number of iterations, 10, is set based on the computing power of the galvanometer controller and the allowable downtime for recalibration. By testing the time consumption of a single iteration and the maximum idle time allowed in the production cycle, it is determined that recalibration can be completed within the maximum number of iterations, thus avoiding the impact of iteration calculation timeout on processing efficiency.

[0026] refer to Figure 4 As shown, a galvanometer scanning positioning accuracy adaptive correction system is used to implement the above-mentioned adaptive calibration method, specifically including: The optical path guidance and calculation module is used to guide the deflection of the probe beam in response to theoretical coordinate control commands. After beam splitting and extraction, the beam spot signal is obtained through the position detection device, and the signal is denoised and centroid calculated to output the measured position signal. The error decoupling calibration module is used to compare the measured position signal with the theoretical coordinates to generate an error dataset, and then decouple it into low-order geometric error and high-order optical distortion according to the physical source. The low-order and high-order correction parameters are generated by matrix solving and spatial interpolation respectively, and the correction parameter set is merged and output. The instruction compensation monitoring module is used to superimpose the set of correction parameters into the real-time control instruction to correct the deflection angle, and control the probe beam to be projected onto the reference point in the non-processing area during the processing gap, and extract the positioning deviation representing the current system drift state by comparison. The adaptive recalibration deduction module is used to respond to the positioning deviation triggering the resampling of a new error dataset, and to structurally separate the current correction parameter set into three components: bias, low-order and high-order. The high-order component is locked and iteratively solved using only the bias and low-order components as initial vectors in combination with the new error dataset. This generates an updated correction parameter set to replace the current parameters in order to achieve error closed-loop convergence.

[0027] Preferably, the optical path guidance and resolution module includes: The coordinate establishment unit is used to establish a reference coordinate system based on the working area of ​​the galvanometer scanning, and to delineate a regular dot matrix sequence covering the working area within the reference coordinate system to obtain the coordinate values ​​of each theoretical coordinate point. The deflection control unit is used to send control commands corresponding to each theoretical coordinate point to the galvanometer system in the order of the rule dot matrix sequence, so as to control the probe beam to be deflected to the corresponding theoretical position by the deflection mirror. The beam splitting and extraction unit is used to extract a portion of the beam from the deflected optical path and guide it to the position detection device. The noise reduction calculation unit is used to perform noise reduction processing on the electrical signal output by the position detection device by multiple sampling and averaging, and then use the centroid algorithm to calculate the measured coordinates of the physical position of the laser spot, which is then output as the measured position signal.

[0028] Preferably, the error decoupling calibration module includes: The error comparison unit is used to collect the coordinate deviations corresponding to the measured position signals of all sampling points into a coordinate error dataset. The physical decoupling unit is used to decouple the coordinate error dataset into low-order geometric error components and high-order optical distortion components according to the physical source of the error. The low-order solving unit is used to construct an affine transformation matrix containing a two-axis scaling factor and an orthogonal bias parameter for the low-order geometric error components, and solve the affine transformation matrix by the least squares method to generate low-order correction parameters for compensating for mechanical and electrical offsets. The higher-order interpolation unit is used to construct a distortion grid table covering the scanning area for the residual spatial deviation after deducting the lower-order geometric error components, and to perform spatial interpolation operations on the distortion grid table to generate higher-order correction parameters for compensating for the inherent distortion of the field lens. The parameter merging unit is used to merge the low-order correction parameters and the high-order correction parameters according to a preset hierarchical data structure, and output a set of correction parameters with hierarchical attributes.

[0029] Preferably, the instruction compensation monitoring module includes: The instruction superposition unit is used to load the correction parameter set into the real-time instruction generation stage of the galvanometer controller. Each time a new control instruction is generated, the current theoretical coordinates are superimposed with the compensation values ​​corresponding to the correction parameter set to obtain the corrected actual execution coordinates. The reference projection unit is used to control the probe beam to project at a probe power that does not produce a substantial processing effect onto a preset reference sampling point located in the non-processing area at the edge of the processing area during the idle period when the current processing trajectory has been completed and the next processing trajectory has not yet started. The deviation extraction unit is used to acquire the measured position signal output by the position detection device at the preset reference sampling point, calculate the difference between the measured position and the theoretical coordinates, and obtain the positioning deviation amount representing the current system drift state. The adaptive recalibration inference module includes: The resampling trigger unit is used to respond to the positioning deviation exceeding the threshold, control the detection beam to retrace the sampling points according to the preset sampling point set, obtain the measured position signal of each sampling point through the position detection device, compare the difference between the measured physical coordinates and the theoretical coordinates, and regenerate the coordinate error dataset. The structure stripping unit is used to perform structured analysis on the currently loaded set of correction parameters, dividing it into offset components representing the overall coordinate translation, low-order distortion components representing the scaling and angular deviation of the two axes, and high-order distortion components representing the inherent distortion of the optical system due to non-uniform spatial distribution. The locking iteration unit is used to keep the higher-order distortion components locked during the iterative solution process, treating them as converged intrinsic optical distortion parameters. The offset component and the lower-order distortion components are used as the initial vectors for iteration, and the dimensionality reduction iterative optimization is performed in combination with the newly generated coordinate error dataset to solve for the updated offset component and the updated lower-order distortion components. The parameter merging and overwriting unit is used to merge the updated bias component, the updated low-order distortion component, and the unchanged high-order distortion component into an updated correction parameter set, and write it into the galvanometer control link to replace the current parameters, thereby achieving error closed-loop convergence.

[0030] The advantage of this invention lies in obtaining the positioning deviation by projecting the probe beam onto a preset reference sampling point in the non-processing area at the edge of the processing area during the processing gap. This enables online quantitative diagnosis of the positioning status of the galvanometer system, which can be directly compared with a preset recalibration threshold. This provides an objective and quantifiable basis for triggering recalibration, avoiding the blindness of relying on manual experience or triggering recalibration at fixed time periods in the prior art. It prevents the decrease in processing accuracy caused by continuing to use the calibration parameters after they have drifted significantly, and also avoids the loss of production efficiency caused by unnecessary recalibration when the parameters are still valid.

[0031] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for adaptive correction of galvanometer scanning positioning accuracy, characterized in that, Specifically, it includes: S1. Send control commands corresponding to the theoretical coordinates to the galvanometer system, rely on the detection beam to be deflected by the deflection mirror, extract part of the beam from the deflected optical path through the beam splitter and guide it to the position detection device, and obtain the measured position signal output by the position detection device corresponding to the physical position of the laser spot. S2. Compare the measured position signal with the theoretical coordinates to obtain a coordinate error dataset, input the coordinate error dataset into the compensation algorithm, calculate and output a set of correction parameters for correcting the control command; S3. Load the correction parameter set into the galvanometer control link, correct the deflection angle of the subsequent control commands, and during the processing after the correction parameter set is loaded, control the probe beam to be projected onto the preset reference sampling point located in the non-processing area at the edge of the processing area to generate the positioning deviation. S4. When the positioning deviation exceeds the threshold, control the probe beam to traverse the regular dot matrix sequence within the processing area and re-traverse the sampling points to generate a coordinate error dataset. Separate the currently loaded correction parameter set into offset components, low-order distortion components and high-order distortion components. S5. Extract the bias component and low-order distortion component as initial parameters for iteration, and keep the high-order distortion component unchanged. Perform iterative solution based on the initial parameters for iteration and the regenerated coordinate error dataset to generate an updated correction parameter set and perform loading.

2. The adaptive correction method for galvanometer scanning positioning accuracy according to claim 1, characterized in that, Step S1 specifically includes: A reference coordinate system is established based on the working area of ​​the galvanometer scanning, and a regular point matrix sequence covering the working area is delineated within the reference coordinate system. The coordinate values ​​of each theoretical coordinate point in the regular point matrix sequence are then obtained. According to the order of the rule lattice sequence, control commands corresponding to each theoretical coordinate point are sequentially sent to the galvanometer system to control the probe beam to be deflected to the corresponding theoretical position by the deflection mirror; Extract a portion of the beam from the deflected optical path and guide it to the position detection device; After performing noise reduction processing by averaging multiple samples on the electrical signal output by the position detection device, the centroid algorithm is used to calculate the measured coordinates of the physical position of the laser spot, which are then used as the measured position signal.

3. The adaptive correction method for galvanometer scanning positioning accuracy according to claim 2, characterized in that, Step S2 specifically includes: The coordinate deviation set corresponding to the measured position signals of all sampling points is called the coordinate error dataset. The coordinate error dataset is decoupled into low-order geometric error components and high-order optical distortion components according to the physical source of the error. For the low-order geometric error components, an affine transformation matrix containing two-axis scaling factors and orthogonal bias parameters is constructed. The affine transformation matrix is ​​solved by the least squares method to generate low-order correction parameters for compensating for mechanical and electrical offsets. For the higher-order optical distortion components, a distortion grid table covering the scanning area is constructed for the residual spatial deviation after deducting the lower-order geometric error components. Spatial interpolation is performed on the distortion grid table to generate higher-order correction parameters for compensating for the inherent distortion of the field lens. The low-order correction parameters and the high-order correction parameters are merged according to a preset data structure to output a set of correction parameters with hierarchical attributes.

4. The adaptive correction method for galvanometer scanning positioning accuracy according to claim 3, characterized in that, Step S3 specifically includes: The set of correction parameters is loaded into the real-time command generation stage of the galvanometer controller. Each time a new control command is generated, the current theoretical coordinates are superimposed with the compensation values ​​corresponding to the set of correction parameters to obtain the corrected actual execution coordinates. After the set of correction parameters is successfully loaded and enters the processing process, during the idle period when the current processing trajectory is completed and the next processing trajectory has not yet started, the probe beam is controlled to be projected at a probe power that does not produce a substantial processing effect onto at least one preset reference sampling point located at the edge of the set processing area and outside the area occupied by the current actual processing graphic. The measured position signal output by the position detection device at the preset reference sampling point is obtained, and the difference between the two-dimensional coordinates corresponding to the measured position and the theoretical two-dimensional coordinates of the preset reference sampling point is calculated to obtain the two-dimensional positioning deviation, which is used as the positioning deviation representing the current positioning state of the galvanometer system.

5. The adaptive correction method for galvanometer scanning positioning accuracy according to claim 4, characterized in that, Step S4 specifically includes: When the positioning deviation exceeds the preset threshold, i.e. the recalibration trigger condition is met, the probe beam is controlled to be projected sequentially onto each sampling point in the processing area according to the preset sampling point set. The sampling point set is completely consistent with the regular dot matrix sequence used when the correction parameter set was initially generated, or a simplified dot matrix sequence formed by selecting intervals of the regular dot matrix sequence to reduce data acquisition time. The measured position signal corresponding to each sampling point is obtained by the position detection device, and the difference between the physical coordinates corresponding to the measured position signal and the theoretical coordinates is compared to regenerate the coordinate error dataset in the current state. The currently loaded set of correction parameters is structured and analyzed based on the physical error sources and spatial variation characteristics corresponding to each parameter item in the set of correction parameters. It is divided into offset components that represent the overall translation of the coordinates, low-order distortion components that represent the scaling and angular deviation of the two axes, and high-order distortion components that represent the inherent distortion of the optical system due to non-uniform spatial distribution.

6. The adaptive correction method for galvanometer scanning positioning accuracy according to claim 5, characterized in that, Step S5 specifically includes: The bias component and the low-order distortion component are extracted as the initial parameter vector for iterative solution, wherein the bias component provides an initial estimate of the current system zero-point translation state, and the low-order distortion component provides an initial estimate of the current system two-axis scaling factor and orthogonal deflection state. The higher-order distortion components are kept locked during this iterative solution process and are regarded as converged intrinsic optical distortion parameters, and are not included in the parameter set to be optimized. Using the regenerated coordinate error dataset as the observations in the objective function, and taking the initial parameter vector as the search starting point, a preset iterative optimization algorithm is executed to solve for the update bias component and update low-order distortion component that minimize the residual. The updated bias component, the updated low-order distortion component, and the unchanged high-order distortion component are merged to generate the updated correction parameter set, and the updated correction parameter set is written into the galvanometer control link to replace the currently loaded correction parameter set.

7. A galvanometer scanning positioning accuracy adaptive correction system, used to implement the adaptive correction method as described in any one of claims 1-6, characterized in that, Specifically, it includes: The optical path guidance and calculation module is used to guide the deflection of the probe beam in response to theoretical coordinate control commands. After beam splitting and extraction, the beam spot signal is obtained through the position detection device, and the signal is denoised and centroid calculated to output the measured position signal. The error decoupling calibration module is used to compare the measured position signal with the theoretical coordinates to generate an error dataset, and then decouple it into low-order geometric error and high-order optical distortion according to the physical source. The low-order and high-order correction parameters are generated by matrix solving and spatial interpolation respectively, and the correction parameter set is merged and output. The instruction compensation monitoring module is used to superimpose the set of correction parameters into the real-time control instruction to correct the deflection angle, and control the probe beam to be projected onto the reference point in the non-processing area during the processing gap, and extract the positioning deviation representing the current system drift state by comparison. The adaptive recalibration deduction module is used to respond to the positioning deviation triggering the resampling of a new error dataset, and to structurally separate the current correction parameter set into three components: bias, low-order and high-order. The high-order component is locked and iteratively solved using only the bias and low-order components as initial vectors in combination with the new error dataset. This generates an updated correction parameter set to replace the current parameters in order to achieve error closed-loop convergence.

8. The adaptive correction system for galvanometer scanning positioning accuracy according to claim 7, characterized in that, The optical path guidance and calculation module includes: The coordinate establishment unit is used to establish a reference coordinate system based on the working area of ​​the galvanometer scanning, and to delineate a regular dot matrix sequence covering the working area within the reference coordinate system to obtain the coordinate values ​​of each theoretical coordinate point. The deflection control unit is used to send control commands corresponding to each theoretical coordinate point to the galvanometer system in the order of the rule dot matrix sequence, so as to control the probe beam to be deflected to the corresponding theoretical position by the deflection mirror. The beam splitting and extraction unit is used to extract a portion of the beam from the deflected optical path and guide it to the position detection device. The noise reduction calculation unit is used to perform noise reduction processing on the electrical signal output by the position detection device by multiple sampling and averaging, and then use the centroid algorithm to calculate the measured coordinates of the physical position of the laser spot, which is then output as the measured position signal.

9. The adaptive correction system for galvanometer scanning positioning accuracy according to claim 7, characterized in that, The error decoupling calibration module includes: The error comparison unit is used to collect the coordinate deviations corresponding to the measured position signals of all sampling points into a coordinate error dataset. The physical decoupling unit is used to decouple the coordinate error dataset into low-order geometric error components and high-order optical distortion components according to the physical source of the error. The low-order solving unit is used to construct an affine transformation matrix containing a two-axis scaling factor and an orthogonal bias parameter for the low-order geometric error components, and solve the affine transformation matrix by the least squares method to generate low-order correction parameters for compensating for mechanical and electrical offsets. The higher-order interpolation unit is used to construct a distortion grid table covering the scanning area for the residual spatial deviation after deducting the lower-order geometric error components, and to perform spatial interpolation operations on the distortion grid table to generate higher-order correction parameters for compensating for the inherent distortion of the field lens. The parameter merging unit is used to merge the low-order correction parameters and the high-order correction parameters according to a preset hierarchical data structure, and output a set of correction parameters with hierarchical attributes.

10. The adaptive correction system for galvanometer scanning positioning accuracy according to claim 9, characterized in that, The instruction compensation monitoring module includes: The instruction superposition unit is used to load the correction parameter set into the real-time instruction generation stage of the galvanometer controller. Each time a new control instruction is generated, the current theoretical coordinates are superimposed with the compensation values ​​corresponding to the correction parameter set to obtain the corrected actual execution coordinates. The reference projection unit is used to control the probe beam to project at a probe power that does not produce a substantial processing effect onto a preset reference sampling point located in the non-processing area at the edge of the processing area during the idle period when the current processing trajectory has been completed and the next processing trajectory has not yet started. The deviation extraction unit is used to acquire the measured position signal output by the position detection device at the preset reference sampling point, calculate the difference between the measured position and the theoretical coordinates, and obtain the positioning deviation amount representing the current system drift state. The adaptive recalibration inference module includes: The resampling trigger unit is used to respond to the positioning deviation exceeding the threshold, control the detection beam to retrace the sampling points according to the preset sampling point set, obtain the measured position signal of each sampling point through the position detection device, compare the difference between the measured physical coordinates and the theoretical coordinates, and regenerate the coordinate error dataset. The structure stripping unit is used to perform structured analysis on the currently loaded set of correction parameters, dividing it into offset components representing the overall coordinate translation, low-order distortion components representing the scaling and angular deviation of the two axes, and high-order distortion components representing the inherent distortion of the optical system due to non-uniform spatial distribution. The locking iteration unit is used to keep the higher-order distortion components locked during the iterative solution process, treating them as converged intrinsic optical distortion parameters. The offset component and the lower-order distortion components are used as the initial vectors for iteration, and the dimensionality reduction iterative optimization is performed in combination with the newly generated coordinate error dataset to solve for the updated offset component and the updated lower-order distortion components. The parameter merging and overwriting unit is used to merge the updated bias component, the updated low-order distortion component, and the unchanged high-order distortion component into an updated correction parameter set, and write it into the galvanometer control link to replace the current parameters, thereby achieving error closed-loop convergence.

Images