Adaptive beam quality optimization method, device, equipment and storage medium

Abstract

The present application relates to the technical field of light beam imaging quality optimization, and discloses an adaptive light beam quality optimization method, device, equipment and storage medium; a phase diagram is constructed according to a preset orthogonal polynomial moment function, a preset first weight coefficient, a preset second weight coefficient and a preset normalized peak intensity; the phase diagram is modulated according to a modal residual index, a preset comprehensive covariance, a preset Fourier transform operator and a preset adaptive learning rate, so as to obtain an expected target light beam distribution; the present scheme constructs a centroid correction model through double-point coordinates, polynomial moment functions and the like, and generates a target light beam in combination with modal residuals and iterative modulation; without harsh temperature control and complex calibration, the present scheme reduces drift errors at a suitable distance, and takes into account accuracy, real-time performance and economy, and can be widely applied to high-value manufacturing scenarios, thereby improving yield and reducing costs.

Description

Technical Field

[0001] This invention relates to the field of beam imaging quality optimization technology, and in particular to adaptive beam quality optimization methods, apparatus, devices, and storage media. Background Technology

[0002] The core challenge in long-distance online 5-DOF error measurement lies in the laser beam drift caused by air turbulence, leading to a decrease in measurement accuracy. At a transmission distance of 10 meters, the laser beam drift caused by air turbulence and temperature fluctuations can reach 50 μm, resulting in decreased measurement accuracy over long distances. This drift not only directly reduces measurement accuracy but also further amplifies system errors due to the nonlinear coupling of multi-degree-of-freedom errors, leading to a decrease in yield and a surge in costs in complex processing scenarios such as semiconductor wafer bonding and aero-engine blade grinding. To address the multi-degree-of-freedom error problem caused by beam drift, scholars at home and abroad have proposed a variety of improvement strategies, attempting to achieve the best balance between accuracy, real-time performance, and cost. The dual-frequency laser interferometry method effectively suppresses noise caused by environmental vibration and air turbulence using differential detection technology with orthogonally polarized light, but it has extremely high requirements for temperature control and optical alignment, resulting in high costs and difficult maintenance for industrial applications. Spatial diversity technology is used to improve anti-interference capabilities, achieving an improvement in the accuracy of six-degree-of-freedom measurement, but the system complexity and calibration time limit its large-scale promotion. The five-degree-of-freedom system based on parallel beams has poor reliability due to the instability of beam parallelism. Summary of the Invention

[0003] In order to overcome the shortcomings of the prior art, the present invention aims to provide methods, apparatus, devices and storage media.

[0004] An adaptive beam quality optimization method includes: obtaining first and second spatial coordinates; calculating the centroid position based on a preset normalized scaling factor, a preset activation function, the first spatial coordinates, and the second spatial coordinates; constructing a phase map based on a preset orthogonal polynomial moment function, a preset first weighting coefficient, a preset second weighting coefficient, and a preset normalized peak intensity; constructing a centroid correction model based on the phase map, a preset first sensitivity coefficient, and a preset second sensitivity coefficient; correcting the centroid position based on the centroid correction model to obtain a corrected centroid position; generating a modal residual index based on the corrected centroid position, a preset third sensitivity coefficient, a preset spatial mapping coefficient, and a preset focal length; and modulating the phase map based on the modal residual index, a preset comprehensive covariance, a preset Fourier transform operator, and a preset adaptive learning rate to obtain the desired target beam distribution.

[0005] Further, the step of calculating the centroid position based on a preset normalized scaling factor, a preset activation function, first spatial coordinates, and second spatial coordinates includes: obtaining the total power of the two-dimensional cross-section, the center offset in the first direction, and the center offset in the second direction; constructing a light intensity distribution function based on a preset spot radius, a preset pi, a preset natural exponential function, the total power of the two-dimensional cross-section, the first spatial coordinates, the second spatial coordinates, the center offset in the first direction, and the center offset in the second direction; obtaining local pixel intensity data in the quadrant, and constructing a weighting function based on the natural exponential function, a preset high-intensity region determination threshold, a normalized scaling factor, and the local pixel intensity data in the quadrant; constructing a binary mask matrix based on the light intensity distribution function, the high-intensity region determination threshold, the normalized scaling factor, and the activation function; and performing a weighted average calculation on the first spatial coordinates and the second spatial coordinates based on the light intensity distribution function, the weighting function, and the binary mask matrix to obtain the centroid position.

[0006] Further, the step of constructing the centroid correction model based on the phase diagram, a preset first sensitivity coefficient, and a preset second sensitivity coefficient includes: generating a first correction factor based on the first sensitivity coefficient and the phase diagram; generating a second correction factor based on the second sensitivity coefficient and the phase diagram; obtaining the photocurrent values ​​in the first quadrant, the second quadrant, the third quadrant, and the fourth quadrant; and constructing the centroid correction model based on the phase diagram, the first correction factor, the second correction factor, the photocurrent values ​​in the first quadrant, the second quadrant, the third quadrant, and the fourth quadrant.

[0007] Further, the step of generating modal residual indices based on the corrected centroid position, a preset third sensitivity coefficient, a preset spatial mapping coefficient, and a preset focal length includes: calculating the photocurrents in the first, second, third, and fourth quadrants based on the second sensitivity coefficient to obtain a beam offset in the first direction; calculating the beam offset in the first direction based on the focal length to obtain a first-direction angular drift error; calculating the photocurrents in the first, second, third, and fourth quadrants based on the third sensitivity coefficient to obtain a beam offset in the second direction; and calculating the beam offset in the second direction based on the focal length. The calculation is performed to obtain the second-direction angular drift error; an angular offset vector is generated based on the first-direction angular drift error and the second-direction angular drift error; an equivalent translational drift vector is generated based on the angular offset vector and a preset propagation distance; a preset fourth sensitivity coefficient, first quadrant photocurrent, second quadrant photocurrent, third quadrant photocurrent, and fourth quadrant photocurrent are calculated to obtain the third-direction beam offset; a translational drift vector is generated based on the first-direction beam offset and the third-direction beam offset; a spatial displacement vector is generated based on the corrected centroid position and spatial mapping coefficient; and a modal residual index is constructed based on the equivalent translational drift vector, the translational drift vector, and the spatial displacement vector.

[0008] Furthermore, the step of modulating the phase map according to the modal residual index, a preset comprehensive covariance, a preset Fourier transform operator, and a preset adaptive learning rate to obtain the desired target beam distribution includes: generating a smoothing response mechanism based on the modal residual index, comprehensive covariance, a preset equivalent drift vector, and a preset weight adjustment factor; constructing a multi-objective loss function based on the Fourier transform operator, preset system requirements, a preset set of adjustment weight coefficients, and the smoothing response mechanism; iteratively optimizing the phase map according to a preset iterative update formula and the multi-objective loss function to obtain a structural phase map; generating a modulated phase map based on the adaptive learning rate, the smoothing response mechanism, a preset differential gain parameter, and a preset partial derivative operator; generating a composite phase map based on the structural phase map, the modulated phase map, and a preset basic reference phase map; and modulating the composite phase map according to preset Fourier optical principles to obtain the desired target beam distribution.

[0009] Furthermore, the construction of the multi-objective loss function based on the Fourier transform operator, preset system requirements, preset set of adjustment weight coefficients, and smoothing response mechanism includes: performing Fourier light field propagation mapping on the phase map based on the Fourier transform operator, preset amplitude distribution function, and preset complex phase factor to obtain a reconstructed phase map; generating a target image according to system requirements; extracting features from the reconstructed phase map to obtain the light spot image gradient; calculating the structural error between the reconstructed phase map and the target image; and constructing the multi-objective loss function based on the reconstructed phase map, structural error, preset absolute norm, preset light spot aspect ratio, light spot image gradient, set of adjustment weight coefficients, and smoothing response mechanism.

[0010] Furthermore, the smoothness response mechanism generated based on the modal residual index, comprehensive covariance, preset equivalent drift vector, and preset weight adjustment factor includes: constructing a channel weight calculation mechanism based on the comprehensive covariance, weight adjustment factor, preset channel drift measurement variance within the sliding window, preset regularization factor, preset residual adjustment weight, preset structural reliability, modal residual index, and preset channel identifier; performing normalized weighted calculation on the equivalent drift vector and channel weight calculation mechanism to obtain a normalized weighted sum; and generating a smoothness response mechanism based on preset update coefficients and the normalized weighted sum.

[0011] Furthermore, the adaptive beam quality optimization device includes: a coordinate acquisition module for acquiring first and second spatial coordinates; a centroid position calculation module for calculating the centroid position based on a preset normalized scaling factor, a preset activation function, the first spatial coordinates, and the second spatial coordinates; a phase map construction module for constructing a phase map based on a preset orthogonal polynomial moment function, a preset first weighting coefficient, a preset second weighting coefficient, and a preset normalized peak intensity; a model construction module for constructing a centroid correction model based on the phase map, a preset first sensitivity coefficient, and a preset second sensitivity coefficient; a correction module for correcting the centroid position based on the centroid correction model to obtain a corrected centroid position; an index generation module for generating a modal residual index based on the corrected centroid position, a preset third sensitivity coefficient, a preset spatial mapping coefficient, and a preset focal length; and a phase map modulation module for modulating the phase map based on the modal residual index, a preset comprehensive covariance, a preset Fourier transform operator, and a preset adaptive learning rate to obtain the desired target beam distribution.

[0012] Furthermore, an adaptive beam quality optimization device includes: a memory and at least one processor, the memory storing instructions; at least one processor invokes the instructions in the memory to cause the adaptive beam quality optimization device to perform the steps of the adaptive beam quality optimization method as described in any one of the above descriptions.

[0013] Furthermore, a computer-readable storage medium stores instructions that, when executed by a processor, implement the steps of the adaptive beam quality optimization method as described in any of the preceding claims.

[0014] In the technical solution of this invention, the centroid position is calculated based on two-point spatial coordinates, combined with a normalized scaling factor and an activation function, providing a benchmark for correction. A phase diagram is constructed using orthogonal polynomial moment functions and weighting coefficients to characterize the multi-degree-of-freedom error coupling relationship. A centroid correction model is established using the phase diagram and sensitivity coefficients to reduce the impact of beam drift. The degree of error convergence is evaluated using modal residual indices, and the desired target beam distribution is generated by combining comprehensive covariance, Fourier transform operators, and adaptive learning rate iterative modulation of the phase diagram. This solution reduces drift error within a suitable distance without requiring stringent temperature control, complex calibration, or high-cost optical arrays, balancing accuracy, real-time performance, and economy. It can be widely deployed in high-value manufacturing scenarios such as wafer bonding and blade grinding, improving product yield and reducing overall costs. Attached Figure Description

[0015] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0016] Figure 1 This is a first flowchart of the adaptive beam quality optimization method provided in an embodiment of the present invention;

[0017] Figure 2 This is a second flowchart of the adaptive beam quality optimization method provided in an embodiment of the present invention;

[0018] Figure 3 This is a third flowchart of the adaptive beam quality optimization method provided in an embodiment of the present invention;

[0019] Figure 4 This is a fourth flowchart of the adaptive beam quality optimization method provided in the embodiments of the present invention;

[0020] Figure 5 This is a fifth flowchart of the adaptive beam quality optimization method provided in an embodiment of the present invention;

[0021] Figure 6The sixth flowchart of the adaptive beam quality optimization method provided in the embodiments of the present invention;

[0022] Figure 7 The seventh flowchart of the adaptive beam quality optimization method provided in the embodiments of the present invention;

[0023] Figure 8 This is a schematic diagram of the adaptive beam quality optimization device provided in an embodiment of the present invention;

[0024] Figure 9 This is a schematic diagram of the adaptive beam quality optimization device provided in an embodiment of the present invention. Detailed Implementation

[0025] The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0026] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 One embodiment of the adaptive beam quality optimization method in this invention includes:

[0027] 101. Obtain the first spatial coordinates and the second spatial coordinates;

[0028] 102. The centroid position is calculated based on the preset normalization scaling factor, the preset activation function, the first spatial coordinates, and the second spatial coordinates;

[0029] In this embodiment, by acquiring the spatial coordinates of two points in the beam's transmission path, and combining the normalized scaling factor (to eliminate the influence of distance on the coordinates) and the activation function (to enhance nonlinear feature extraction), the centroid position is calculated, the macroscopic trend of beam drift is quantified, and a benchmark is provided for subsequent correction.

[0030] 103. A phase diagram is constructed based on the preset orthogonal polynomial moment function, the preset first weight coefficient, the preset second weight coefficient, and the preset normalized peak intensity.

[0031] In this embodiment, the expression for constructing the phase map is as follows:

[0032] (1)

[0033] In the formula, The first weighting coefficient, The second weighting coefficient is... Orthogonal polynomial moment functions To normalize the peak intensity, This is a phase diagram; this phase diagram can simultaneously characterize the coupling relationship of multiple degrees of freedom errors such as beam translation and rotation, breaking through the limitation of traditional methods that are difficult to decouple nonlinear errors;

[0034] 104. Based on the phase diagram, the preset first sensitivity coefficient, and the preset second sensitivity coefficient, a centroid correction model is constructed.

[0035] 105. Correct the position of the centroid based on the centroid correction model to obtain the corrected centroid position;

[0036] 106. Generate modal residual indices based on the corrected centroid position, the preset third sensitivity coefficient, the preset spatial mapping coefficient, and the preset focal length;

[0037] In this embodiment, the degree of convergence of system error is evaluated by constructing a modal residual index;

[0038] 107. Modulate the phase map according to the modal residual index, the preset comprehensive covariance, the preset Fourier transform operator, and the preset adaptive learning rate to obtain the desired target beam distribution.

[0039] In this embodiment, the phase map is iteratively modulated using modal residual index, comprehensive covariance (statistical error distribution), Fourier transform operator (frequency domain filtering and noise reduction) and adaptive learning rate (dynamically adjusting the optimization step size) to finally generate the desired target beam distribution and achieve error suppression.

[0040] In this embodiment, the centroid position is calculated based on two-point spatial coordinates, combined with a normalized scaling factor and activation function, providing a benchmark for correction. A phase diagram is constructed using orthogonal polynomial moment functions and weighting coefficients to characterize the multi-degree-of-freedom error coupling relationship. A centroid correction model is established using the phase diagram and sensitivity coefficients to reduce the impact of beam drift. The degree of error convergence is evaluated using modal residual indices, and the desired target beam distribution is generated by iteratively modulating the phase diagram using comprehensive covariance, Fourier transform operators, and adaptive learning rate. This solution reduces drift error within a suitable distance without requiring stringent temperature control, complex calibration, or high-cost optical arrays, balancing accuracy, real-time performance, and economy. It can be widely deployed in high-value manufacturing scenarios such as wafer bonding and blade grinding, improving product yield and reducing overall costs.

[0041] Please see Figure 2 The second embodiment of the adaptive beam quality optimization method in this invention includes:

[0042] 201. Obtain the total power of the two-dimensional cross-section, the center offset in the first direction, and the center offset in the second direction;

[0043] 202. A light intensity distribution function is constructed based on the preset spot radius, preset pi, preset natural exponential function, total power of the two-dimensional cross section, first spatial coordinates, second spatial coordinates, first direction center offset, and second direction center offset.

[0044] In this embodiment, the expression for the light intensity distribution function is:

[0045] (2)

[0046] In the formula, Let ω represent the total power of the two-dimensional cross section, ω represent the beam radius, defined as the radius corresponding to the beam intensity decreasing to 1 / e2 of its maximum value, exp(·) represent the natural exponential function, Δx and Δy represent the center drift of the beam in the x and y directions, respectively (i.e., the center offset in the first direction and the center offset in the second direction); (x, y) are the spatial coordinates on the QPD receiving surface. For the original CCD image at position The light intensity value at that location;

[0047] 203. Obtain local pixel intensity data in the quadrant, and construct a weight function based on the natural exponential function, the preset high-intensity region judgment threshold, the normalized scale factor, and the local pixel intensity data in the quadrant;

[0048] In this embodiment, the expression for the weighting function is:

[0049] (3)

[0050] In the formula, This represents the intensity value of a local pixel in the i-th quadrant of a CCD image. This is the normalized scaling factor that controls the slope of the function; For weighting functions;

[0051] 204. A binary mask matrix is ​​constructed based on the light intensity distribution function, the high-intensity region determination threshold, the normalized scale factor, and the activation function;

[0052] In this embodiment, the expression for the binary mask matrix is:

[0053] (4)

[0054] In the formula, The threshold for determining high-intensity regions; It is a binary mask matrix;

[0055] It is a commonly used "S-shaped" function, whose core function is to map any real number input to the interval (0, 1);

[0056] 205. Calculate the centroid position by performing a weighted average of the first and second spatial coordinates based on the light intensity distribution function, weighting function, and binary mask matrix;

[0057] In this embodiment, the final centroid position is calculated by the weighted average of I(x,y), and the specific formula is as follows:

[0058] (5)

[0059] To extend the response range of the QPD in large offset scenes, when the spot deviates from the linear operating region of the QPD, causing a sharp drop in sensitivity, the geometric center of the spot output in real time from the CCD is used. For reference, the drift mapping relationship between QPD and CCD is established by combining the QPD output values ​​ΔX and ΔY through nonlinear polynomial fitting, thereby correcting the drift distortion of QPD in the boundary region;

[0060] In this embodiment, a light intensity distribution function is constructed based on the total power of the two-dimensional cross section, the spot radius, and the center offset to characterize the spot energy distribution. A weighting function is designed in combination with the local pixel intensity data of the quadrant. By constructing a binary mask matrix, the contribution of high-intensity regions to the centroid calculation is highlighted, and noise interference is suppressed. By fusing the light intensity distribution and spatial coordinate information through weighted averaging, the centroid positioning accuracy is improved, and the stability under large offset scenarios is enhanced. In long-distance laser measurement, the system's anti-interference capability can be enhanced, providing reliable centroid detection support for precision manufacturing, aerospace, and other fields.

[0061] Please see Figure 3 The third embodiment of the adaptive beam quality optimization method in this invention includes:

[0062] 301. Generate the first correction factor based on the first sensitivity coefficient and the phase diagram;

[0063] In this embodiment, Here, (Φ) represents the first sensitivity coefficient, and (Φ) represents the phase diagram. The first correction factor;

[0064] 302. Generate a second correction factor based on the second sensitivity coefficient and the phase diagram;

[0065] In this embodiment, Here, (Φ) represents the second sensitivity coefficient, and (Φ) represents the phase diagram. The second correction factor;

[0066] 303. Obtain the photocurrent values ​​in the first quadrant, the second quadrant, the third quadrant, and the fourth quadrant;

[0067] 304. Based on the phase diagram, the first correction factor, the second correction factor, the photocurrent values ​​in the first quadrant, the photocurrent values ​​in the second quadrant, the photocurrent values ​​in the third quadrant, and the photocurrent values ​​in the fourth quadrant, a centroid correction model is constructed.

[0068] In this embodiment, the expression for the centroid correction model is as follows:

[0069] (6) Correct the x-coordinate of the centroid. Correct the centroid ordinate. The photocurrent value in the first quadrant (the output photocurrent in the first quadrant of the photodetector). The second quadrant photocurrent value (the output photocurrent in the second quadrant of the photodetector) The photocurrent value in the third quadrant (the output photocurrent in the third quadrant of the photodetector) and Fourth quadrant photocurrent value (the output photocurrent in the fourth quadrant of the photodetector);

[0070] In this embodiment, the centroid of the laser beam is calibrated by constructing a centroid correction model, which effectively solves the measurement error problem caused by beam drift. The core is to use the phase diagram and sensitivity coefficient to generate the first correction factor and the second correction factor, and combine the four-quadrant photocurrent values ​​to dynamically correct the centroid coordinates. The correction model integrates phase characteristics and photoelectric detection data, which not only captures the influence of beam wavefront distortion on the centroid, but also achieves dual correction of the centroid's horizontal and vertical coordinates by quantizing the translation offset through quadrant photocurrent differences. This mechanism is suitable for the stringent measurement requirements of semiconductor manufacturing, aerospace and other fields.

[0071] Please see Figure 4 The fourth embodiment of the adaptive beam quality optimization method in this invention includes:

[0072] 401. Calculate the photocurrent in the first quadrant, the photocurrent in the second quadrant, the photocurrent in the third quadrant, and the photocurrent in the fourth quadrant based on the second sensitivity coefficient to obtain the beam offset in the first direction.

[0073] In this embodiment, when the beam undergoes translational drift, ideally, the expression for the beam offset in the first direction is:

[0074] (7)

[0075] In the formula, This is the sensitivity coefficient (second sensitivity coefficient) in the Y-axis direction.

[0076] 402. Calculate the beam offset in the first direction based on the focal length to obtain the angle drift error in the first direction;

[0077] In this embodiment, when the system experiences angular drift, the laser beam will deflect, causing a shift in the position of the spot on the PSD. The angular drift error in the Z-axis direction can be calculated using the following formula:

[0078] (8),

[0079] In the formula, Let f represent the angular drift in the Y direction, and let f be the focal length of L. This indicates the change in the position of the light spot on the PSD in the Y direction (first direction angular drift error).

[0080] 403. Calculate the photocurrent in the first quadrant, the photocurrent in the second quadrant, the photocurrent in the third quadrant, and the photocurrent in the fourth quadrant based on the third sensitivity coefficient to obtain the beam offset in the second direction.

[0081] In this embodiment, the expression for the second direction beam offset is:

[0082] (9)

[0083] In the formula, This is the sensitivity coefficient (third sensitivity coefficient) in the Z-axis direction.

[0084] 404. Calculate the beam offset in the second direction based on the focal length to obtain the angular drift error in the second direction;

[0085] In this embodiment, when the system experiences angular drift, the laser beam will deflect, causing a shift in the position of the spot on the PSD. The angular drift error in the Z-axis direction can be calculated using the following formula:

[0086] (10)

[0087] In the formula, Indicates the angular drift in the Z direction. This indicates the change in the position of the light spot on the PSD in the X direction (second direction angular drift error).

[0088] 405. Generate an angular offset vector based on the first direction angular drift error and the second direction angular drift error;

[0089] In this embodiment, the PSD4 sensor is used to detect laser beam angle drift and outputs an angular offset vector. The expression for the angular offset vector is as follows:

[0090] (11)

[0091] Let t be the angle drift vector, and t be the time window.

[0092] 406. Generate an equivalent translation drift vector based on the angular offset vector and the preset propagation distance;

[0093] In this embodiment, to achieve a unified dimensional conversion of different modal drift amounts, the angular drift information is first converted into an equivalent translational drift vector under the propagation distance L through geometric mapping. The expression for the equivalent translational drift vector is as follows:

[0094] (12);

[0095] In the formula, For transmission distance, This is the equivalent translation drift vector;

[0096] 407. Calculate the preset fourth sensitivity coefficient, first quadrant photocurrent, second quadrant photocurrent, third quadrant photocurrent and fourth quadrant photocurrent to obtain the third-direction beam offset.

[0097] 408. Generate a translational drift vector based on the beam offset in the first direction and the beam offset in the third direction;

[0098] In this embodiment, the QPD sensor provides a translational drift vector, the expression of which is as follows:

[0099] (13);

[0100] In the formula, The translation drift vector, This represents the beam offset in the first direction. This represents the offset of the third-direction beam.

[0101] 409. Generate a spatial displacement vector based on the corrected centroid position and spatial mapping coefficients;

[0102] In this embodiment, the CCD centroid coordinates are simultaneously converted into a spatial displacement vector using the pixel-space mapping coefficient SCCD, thereby achieving a consistent representation of the drift of all channels in the spatial coordinate system. The expression for the spatial displacement vector is as follows:

[0103] (14);

[0104] In the formula, It is a spatial displacement vector. For spatial mapping coefficients, For CCD cameras, the position of the center of mass of the light spot is used (correcting the center of mass position). Output drift coordinates ;

[0105] 410. Modal residual indices are constructed based on the equivalent translational drift vector, translational drift vector, and spatial displacement vector;

[0106] In this embodiment, a modal residual index is constructed to evaluate the drift consistency between channels. The expression for the modal residual index is as follows:

[0107] (15);

[0108] In the formula, when a channel deviates significantly from the observations of the other two channels, β(t) will increase, triggering a dynamic decay of the channel's weight:

[0109] In this embodiment, the core lies in decoupling angular drift and translational drift using quadrant photocurrent: first, the current is mapped to the offset of each beam using a sensitivity coefficient, then the angular drift error is calculated using focal length conversion, generating an angular offset vector; subsequently, the angular drift is converted into an equivalent translational drift through geometric mapping, achieving dimensional unification; the QPD synchronously outputs the true translational vector, while the CCD outputs the spatial displacement vector using spatial mapping coefficients. Finally, the consistency of the three channels is dynamically quantified through modal residual index; when a channel deviates significantly, its weight is automatically reduced to suppress the influence of abnormal data. This solution integrates electrical domain sensitivity calibration, optical focal length conversion, and geometric mapping compensation, improving the stability and reliability of the system under complex environments such as vibration and temperature drift. It is suitable for precision laser communication, processing, and measurement fields and has engineering promotion value.

[0110] Please see Figure 5 The fifth embodiment of the adaptive beam quality optimization method in this invention includes:

[0111] 501. Generate a smoothness response mechanism based on the comprehensive covariance, the preset equivalent drift vector, and the preset weight adjustment factor;

[0112] 502. A multi-objective loss function is constructed based on the Fourier transform operator, the preset system requirements, the preset set of adjustment weight coefficients, and the smoothness response mechanism.

[0113] 503. The phase map is iteratively optimized according to the preset iterative update formula and multi-objective loss function to obtain the structural phase map;

[0114] In this embodiment, the expression for the iterative update formula is as follows:

[0115] (16)

[0116] In the formula, For partial derivative operators, For the structure phase diagram, This is the phase diagram from the previous round. This represents the adaptive learning rate, which dynamically adjusts over time to adapt to different stages of error sensitivity. This path primarily regulates the structure and intensity distribution of the light spot, providing morphological constraints for the final composite phase map. For multi-objective loss functions,

[0117] 504. Generate a control phase map based on the adaptive learning rate, smoothness response mechanism, preset differential gain parameters, and preset partial derivative operators;

[0118] In this embodiment, the expression for the phase map is as follows:

[0119] (17)

[0120] In the formula, This is the phase diagram obtained from the previous iteration, where ∂ represents the partial derivative operator. The differential gain parameter, The adaptive learning rate is used to dynamically adjust the step size according to the error magnitude and trend, effectively avoiding overcompensation or oscillation. The differential term responds in advance to the error trend, improving the ability to respond quickly to disturbances. For phase diagram, It is a norm 2. To regulate the phase diagram, the differential term anticipates the trend of error changes, thereby enhancing the ability to respond quickly to dynamic disturbances. For smoothness response mechanism;

[0121] 505. Generate a composite phase diagram based on the structural phase diagram, the control phase diagram, and the preset basic reference phase diagram;

[0122] In this embodiment, the drift compensation path generates a controlled phase map. Structure phase diagram of beam quality optimization path output and the basic reference phase The layers are superimposed to form a composite phase map loaded onto the SLM;

[0123] (18)

[0124] 506. Modulate the composite phase diagram according to the preset Fourier optical principle to obtain the desired target beam distribution;

[0125] In this embodiment, the composite phase diagram serves as the core for controlling the system's output wavefront. After being loaded onto the SLM, it modulates the incident beam, causing it to be reconstructed into the desired target beam distribution according to Fourier optics principles during its propagation in free space. The desired target beam distribution is expressed as follows:

[0126] (19)

[0127] In the formula, For the complex amplitude distribution of the output light field, Here, is the Fourier transform operator, A(x,y) is the amplitude distribution function of the incident light on the SLM surface, j is the imaginary unit, and exp[·] is the natural exponential function.

[0128] In this embodiment, a smoothness response mechanism based on comprehensive covariance and equivalent drift vector is used to balance system stability and response speed. A multi-objective loss function integrates Fourier transform operators and adjustment weights to constrain the beam structure and intensity distribution. The iterative update formula dynamically adapts to error sensitivity through an adaptive learning rate and introduces differential terms to respond to disturbance trends in advance, avoiding overcompensation or oscillation. Iterative optimization yields a controlled phase map. Finally, the structural phase map, the controlled phase map, and the basic reference phase are superimposed to form a composite phase map, which is then modulated and reconstructed into the desired beam distribution according to Fourier optics principles. This scheme significantly improves the accuracy of multi-degree-of-freedom measurements while balancing real-time performance and robustness. In semiconductor manufacturing, precision machining, and other scenarios, it can reduce system errors, improve production yield, and reduce costs.

[0129] Please see Figure 6 The sixth embodiment of the adaptive beam quality optimization method in this invention includes:

[0130] 601. Perform Fourier light field propagation mapping on the phase map according to the Fourier transform operator, the preset amplitude distribution function, and the preset complex phase factor to obtain the reconstructed phase map.

[0131] In this embodiment, the expression for reconstructing the phase map is as follows:

[0132] (20)

[0133] In the formula, To reconstruct the phase map, Let be the Fourier transform operator, representing the propagation of the light field from the SLM surface to the far field, and let A(x,y) be the amplitude distribution function of the incident light on the SLM surface. Let j be the complex phase factor, j be the imaginary unit, and (x,y) be the spatial coordinates on the QPD contact surface.

[0134] 602. Generate the target image according to system requirements;

[0135] In this embodiment, the target image It can be preset to Gaussian, ring, or mission-specific distribution according to system requirements, and dynamically updated using an exponential moving average strategy to adapt to beam quality changes caused by complex environmental disturbances.

[0136] 603. Extract features from the reconstructed phase map to obtain the gradient of the light spot image;

[0137] 604. Calculate the reconstructed phase map and target image to obtain the structural error;

[0138] 605. A multi-objective loss function is constructed based on the reconstructed phase map, structural error, preset absolute norm, preset spot aspect ratio, spot image gradient, adjusted weight coefficient set, and smoothness response mechanism.

[0139] In this embodiment, the expression for the multi-objective loss function is as follows:

[0140] (twenty one)

[0141] In the formula, the first weighting coefficient Second weighting coefficient Third weighting coefficient Fourth weighting coefficient Adjust the weights for each sub-target (forming a set of adjustment weight coefficients), where a / b represents the ratio of the major and minor axes of the light spot, used to control the ellipticity. For light spot image The gradient represents the degree of change in image edges; Represents the target image With reconstructed images Structural errors between them; It is the absolute value norm, used to measure image symmetry and edge smoothness;

[0142] In this embodiment, the phase map is reconstructed through Fourier light field propagation mapping, and combined with dynamically updated target images (Gaussian, ring, etc. distributions) to achieve precise control of beam characteristics. The multi-target loss function fuses structural error, beam gradient, aspect ratio, and absolute norm, and balances each index through a weighted coefficient set, ensuring both the consistency of beam shape with the target and controlling ellipticity and edge smoothness. The exponential moving average strategy dynamically adapts to environmental disturbances, and the first-order recursive filtering balances smoothness and response speed. The overall mechanism effectively suppresses beam distortion caused by air turbulence, improves the stability of light field distribution in long-distance measurements, and adapts to various scenario requirements, balancing real-time performance and reliability in fields such as semiconductor processing and precision manufacturing.

[0143] Please see Figure 7The seventh embodiment of the adaptive beam quality optimization method in this invention includes:

[0144] 701. A channel weighting calculation mechanism is constructed based on the comprehensive covariance, weighting adjustment factor, preset channel drift measurement variance within the sliding window, preset regularization factor, preset residual adjustment weight, preset structural reliability, modal residual index, and preset channel identifier.

[0145] In this embodiment, the expression for the channel weight calculation mechanism is as follows:

[0146] ,(twenty two)

[0147] In the formula, when a channel deviates significantly from the observations of the other two channels, the β(t) modal residual index will increase, triggering a dynamic decay of the channel's weight. This paper introduces a task-oriented weight adjustment factor. To clarify the physical dominance of different channels under different compensation objectives, g∈{P,Q,C} is the channel identifier; Let ε be the variance of drift measurement within the channel sliding window, ε be the regularization factor, and λ be the residual adjustment weight. Indicates structural credibility. To synthesize covariance;

[0148] 702. Perform normalized weighted calculation on the equivalent drift vector and channel weight operation mechanism to obtain a normalized weighted sum;

[0149] In this embodiment, the expression for the normalized weighted sum is as follows:

[0150] ,(twenty three)

[0151] In the formula, For sensors The calculated equivalent drift vector, For a normalized weighted sum;

[0152] 703. Generate a smoothness response mechanism based on preset update coefficients and normalized weighted sums;

[0153] In this embodiment, to make the fusion result more stable and reliable without excessively delaying the system response, a first-order recursive filter is introduced to smooth the perturbation. The expression for the smoothness response mechanism is as follows:

[0154] ,(twenty four)

[0155] In the formula, γ is the update coefficient of the first-order recursive filter, which is used to control the smoothness of the output response;

[0156] In this embodiment, a multi-dimensional channel weighting calculation mechanism is constructed to suppress beam drift in long-distance laser measurement. The core of this mechanism is the dynamic adjustment of channel weights: when the observation of a certain channel deviates significantly, the modal residual index rises, triggering weight decay. Combined with task-oriented adjustment factors, the physical dominance of each channel is clarified. At the same time, the measurement variance and structural reliability are balanced by regularization factors and residual adjustment weights to solve the problem of multi-channel data conflict. Normalized weighted calculations fuse the equivalent drift vectors of multiple sensors. First-order recursive filtering achieves a balance between smoothing disturbances and system response speed by updating coefficients, avoiding excessive delay. The overall mechanism effectively decouples the nonlinear error caused by air turbulence, improves the measurement stability in complex environments, and adapts to different compensation targets, taking into account the real-time and reliability requirements of industrial scenarios.

[0157] The adaptive beam quality optimization method in the embodiments of the present invention has been described above. The adaptive beam quality optimization device in the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 8 One embodiment of the adaptive beam quality optimization device in this invention includes:

[0158] Coordinate acquisition module 1 is used to acquire first spatial coordinates and second spatial coordinates;

[0159] The centroid position calculation module 2 is used to calculate the centroid position based on a preset normalization scale factor, a preset activation function, first spatial coordinates, and second spatial coordinates.

[0160] Phase diagram construction module 3 is used to construct a phase diagram based on a preset orthogonal polynomial moment function, a preset first weighting coefficient, a preset second weighting coefficient, and a preset normalized peak intensity.

[0161] Model building module 4 is used to build a centroid correction model based on the phase diagram, a preset first sensitivity coefficient, and a preset second sensitivity coefficient.

[0162] Correction module 5 is used to correct the position of the centroid according to the centroid correction model to obtain the corrected centroid position;

[0163] The index generation module 6 is used to generate modal residual indices based on the corrected centroid position, the preset third sensitivity coefficient, the preset spatial mapping coefficient, and the preset focal length.

[0164] The phase map modulation module 7 is used to modulate the phase map according to the modal residual index, the preset comprehensive covariance, the preset Fourier transform operator, and the preset adaptive learning rate to obtain the desired target beam distribution.

[0165] In this embodiment, the centroid position is calculated based on two-point spatial coordinates, combined with a normalized scaling factor and activation function, providing a benchmark for correction. A phase diagram is constructed using orthogonal polynomial moment functions and weighting coefficients to characterize the multi-degree-of-freedom error coupling relationship. A centroid correction model is established using the phase diagram and sensitivity coefficients to reduce the impact of beam drift. The degree of error convergence is evaluated using modal residual indices, and the desired target beam distribution is generated by iteratively modulating the phase diagram using comprehensive covariance, Fourier transform operators, and adaptive learning rate. This solution reduces drift error within a suitable distance without requiring stringent temperature control, complex calibration, or high-cost optical arrays, balancing accuracy, real-time performance, and economy. It can be widely deployed in high-value manufacturing scenarios such as wafer bonding and blade grinding, improving product yield and reducing overall costs.

[0166] Figure 9 This is a schematic diagram of the structure of the adaptive beam quality optimization device 900 provided in the embodiments of the present invention. The adaptive beam quality optimization device 900 can vary considerably due to different configurations or performance. It may include one or more central processing units (CPUs) 910 (e.g., one or more processors) and a memory 920, and one or more storage media 930 (e.g., one or more mass storage devices) storing application programs 933 or data 932. The memory 920 and storage media 930 can be temporary or persistent storage. The program stored in the storage media 930 may include one or more modules (not shown in the diagram), each module may include a series of instruction operations on the adaptive beam quality optimization device 900. Furthermore, the processor 910 may be configured to communicate with the storage media 930 and execute the series of instruction operations in the storage media 930 on the adaptive beam quality optimization device 900 to implement the steps of the adaptive beam quality optimization method provided in the above-described method embodiments.

[0167] The adaptive beam quality optimization device 900 may also include one or more power supplies 940, one or more wired or wireless network interfaces 950, one or more input / output interfaces 960, and / or one or more operating systems 931, such as Windows Server, MacOSX, Unix, Linux, FreeBSD, etc. Those skilled in the art will understand that... Figure 9 The illustrated adaptive beam quality optimization device structure does not constitute a limitation on the adaptive beam quality optimization device, which may include more or fewer components than illustrated, or combine certain components, or have different component arrangements.

[0168] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the steps of the adaptive beam quality optimization method.

[0169] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system, device, or unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0170] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0171] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method of adaptive beam quality optimization, characterized in that, The method comprises the following steps: obtaining a first spatial coordinate and a second spatial coordinate; calculating a centroid position according to a preset normalization scale factor, a preset activation function, the first spatial coordinate and the second spatial coordinate; constructing a phase diagram according to a preset orthogonal polynomial moment function, a preset first weight coefficient, a preset second weight coefficient and a preset normalized peak intensity; constructing a centroid correction model according to the phase diagram, a preset first sensitivity coefficient and a preset second sensitivity coefficient; correcting the centroid position according to the centroid correction model to obtain a corrected centroid position; generating a modal residual index according to the corrected centroid position, a preset third sensitivity coefficient, a preset spatial mapping coefficient and a preset focal length; modulating the phase diagram according to the modal residual index, a preset comprehensive covariance, a preset Fourier transform operator and a preset adaptive learning rate to obtain an expected target beam distribution; The method comprises the following steps: obtaining a two-dimensional cross-sectional total power, a first direction center offset and a second direction center offset; constructing an intensity distribution function according to a preset spot radius, a preset circular constant, a preset natural exponential function, the two-dimensional cross-sectional total power, the first spatial coordinate, the second spatial coordinate, the first direction center offset and the second direction center offset; obtaining quadrant local pixel intensity data, and constructing a weight function according to the natural exponential function, a preset high-intensity region determination threshold, a normalization scale factor and the quadrant local pixel intensity data; constructing a binary mask matrix according to the intensity distribution function, the weight function and the binary mask matrix; performing weighted average calculation on the first spatial coordinate and the second spatial coordinate according to the intensity distribution function, the weight function and the binary mask matrix to obtain the centroid position; The method comprises the following steps: generating a first correction factor according to the first sensitivity coefficient and the phase diagram; generating a second correction factor according to the second sensitivity coefficient and the phase diagram; obtaining a first quadrant photocurrent value, a second quadrant photocurrent value, a third quadrant photocurrent value and a fourth quadrant photocurrent value; constructing the centroid correction model according to the phase diagram, the first correction factor, the second correction factor, the first quadrant photocurrent value, the second quadrant photocurrent value, the third quadrant photocurrent value and the fourth quadrant photocurrent value; The method comprises the following steps: calculating the first quadrant photocurrent, the second quadrant photocurrent, the third quadrant photocurrent and the fourth quadrant photocurrent according to the second sensitivity coefficient to obtain a first direction beam offset; calculating the first direction beam offset according to the focal length to obtain a first direction angle drift error; calculating the first quadrant photocurrent, the second quadrant photocurrent, the third quadrant photocurrent and the fourth quadrant photocurrent according to the third sensitivity coefficient to obtain a second direction beam offset; calculating the second direction beam offset according to the focal length to obtain a second direction angle drift error; generating an angle offset vector according to the first direction angle drift error and the second direction angle drift error; generating an equivalent translation drift vector according to the angle offset vector and a preset propagation distance; calculating a preset fourth sensitivity coefficient, a first quadrant photocurrent, a second quadrant photocurrent, a third quadrant photocurrent and a fourth quadrant photocurrent to obtain a third direction beam offset; generating a translation drift vector according to the first direction beam offset and the third direction beam offset; generating a spatial displacement vector according to the corrected centroid position and a spatial mapping coefficient; constructing a modal residual index according to the equivalent translation drift vector, the translation drift vector and the spatial displacement vector.

2. The adaptive beam quality optimization method of claim 1, wherein, modulating the phase map according to the modal residual index, a preset comprehensive covariance, a preset Fourier transform operator, a preset adaptive learning rate to obtain an expected target beam distribution, including: generating a smoothing degree response mechanism according to the modal residual index, the comprehensive covariance, a preset equivalent drift vector and a preset weight adjustment factor; constructing a multi-objective loss function according to the Fourier transform operator, a preset system requirement, a preset adjustment weight coefficient set and the smoothing degree response mechanism; iteratively optimizing the phase map according to a preset iterative update formula and the multi-objective loss function to obtain a structure phase map; generating a regulated phase map according to the adaptive learning rate, the smoothing degree response mechanism, a preset differential gain parameter and a preset partial derivative operator; generating a composite phase map according to the structure phase map, the regulated phase map and a preset basic reference phase map; modulating the composite phase map according to a preset Fourier optics principle to obtain the expected target beam distribution.

3. The adaptive beam quality optimization method of claim 2, wherein, constructing the multi-objective loss function according to the Fourier transform operator, the preset system requirement, the preset adjustment weight coefficient set and the smoothing degree response mechanism, including: performing Fourier light field propagation mapping on the phase map according to the Fourier transform operator and a preset amplitude distribution function, a preset complex phase factor to obtain a reconstructed phase map; generating a target image according to the system requirement; extracting features from the reconstructed phase map to obtain a spot image gradient; calculating the reconstructed phase map and the target image to obtain a structure error; constructing the multi-objective loss function according to the reconstructed phase map, the structure error, a preset absolute norm, a preset spot major and minor axis ratio, the spot image gradient, the adjustment weight coefficient set and the smoothing degree response mechanism.

4. The adaptive beam quality optimization method of claim 2, wherein, generating the smoothing degree response mechanism according to the modal residual index, the comprehensive covariance, a preset equivalent drift vector and a preset weight adjustment factor, including: constructing a channel weight operation mechanism according to the comprehensive covariance, the weight adjustment factor, a preset in-channel sliding window drift measurement variance, a preset regularization factor, a preset residual adjustment weight, a preset structure reliability, the modal residual index and a preset channel identifier; performing normalized weighted calculation on the equivalent drift vector and the channel weight operation mechanism to obtain a normalized weighted sum; generating the smoothing degree response mechanism according to a preset update coefficient and the normalized weighted sum.

5. An adaptive beam quality optimization apparatus, characterized by, including: The coordinate acquisition module is configured to acquire the first spatial coordinate and the second spatial coordinate. The centroid position calculation module is configured to calculate the centroid position according to a preset normalization scale factor, a preset activation function, the first spatial coordinate and the second spatial coordinate, and specifically includes: Obtain the two-dimensional cross-sectional total power, the first direction center offset and the second direction center offset. Construct the light intensity distribution function according to the preset spot radius, the preset constant pi, the preset natural exponential function, the two-dimensional cross-sectional total power, the first spatial coordinate, the second spatial coordinate, the first direction center offset and the second direction center offset. Obtain the quadrant local pixel intensity data, and construct the weight function according to the natural exponential function, a preset high-intensity region determination threshold, the normalization scale factor and the quadrant local pixel intensity data. Construct the binary mask matrix according to the light intensity distribution function, the weight function and the binary mask matrix. Perform weighted average calculation on the first spatial coordinate and the second spatial coordinate according to the light intensity distribution function, the weight function and the binary mask matrix to obtain the centroid position. The phase map construction module is configured to construct the phase map according to a preset orthogonal polynomial matrix function, a preset first weight coefficient, a preset second weight coefficient and a preset normalized peak intensity. The model construction module is configured to construct the centroid correction model according to the phase map, a preset first sensitivity coefficient and a preset second sensitivity coefficient, and specifically includes: Generate the first correction factor according to the first sensitivity coefficient and the phase map. Generate the second correction factor according to the second sensitivity coefficient and the phase map. Obtain the first quadrant photocurrent value, the second quadrant photocurrent value, the third quadrant photocurrent value and the fourth quadrant photocurrent value. Construct the centroid correction model according to the phase map, the first correction factor, the second correction factor, the first quadrant photocurrent value, the second quadrant photocurrent value, the third quadrant photocurrent value and the fourth quadrant photocurrent value. The correction module is configured to correct the centroid position according to the centroid correction model to obtain a corrected centroid position. The index generation module is configured to generate a modal residual index according to the corrected centroid position, a preset third sensitivity coefficient, a preset spatial mapping coefficient and a preset focal length, and specifically includes: Calculate the first direction beam offset according to the second sensitivity coefficient and the first quadrant photocurrent, the second quadrant photocurrent, the third quadrant photocurrent and the fourth quadrant photocurrent. Calculate the first direction angle drift error according to the focal length and the first direction beam offset. Calculate the second direction beam offset according to the third sensitivity coefficient and the first quadrant photocurrent, the second quadrant photocurrent, the third quadrant photocurrent and the fourth quadrant photocurrent. Calculate the second direction angle drift error according to the focal length and the second direction beam offset. Generate an angular offset vector according to the first direction angle drift error and the second direction angle drift error. Generate an equivalent translational drift vector according to the angular offset vector and a preset propagation distance. The fourth sensitivity coefficient, the first quadrant photocurrent, the second quadrant photocurrent, the third quadrant photocurrent and the fourth quadrant photocurrent are calculated to obtain a third direction light beam offset; A translation drift vector is generated according to the first direction light beam offset and the third direction light beam offset; A spatial displacement vector is generated according to the corrected centroid position and the spatial mapping coefficient; A modal residual index is constructed according to the equivalent translation drift vector, the translation drift vector and the spatial displacement vector; The phase pattern modulation module is configured to modulate the phase pattern according to the modal residual index, a preset comprehensive covariance, a preset Fourier transform operator and a preset adaptive learning rate, so as to obtain an expected target light beam distribution.

6. An adaptive beam quality optimization device, characterized by The adaptive light beam quality optimization device comprises a memory and at least one processor, and the memory stores instructions; The at least one processor invokes the instructions in the memory, so that the adaptive light beam quality optimization device performs the steps of the adaptive light beam quality optimization method according to any one of claims 1-4.

7. A computer-readable storage medium having stored thereon instructions, the computer-readable storage medium comprising: The instructions are executed by the processor to implement the steps of the adaptive light beam quality optimization method according to any one of claims 1-4.

Images