Laser lens processing digital twin system

Abstract

This application provides a digital twin system for laser lens processing, applied in the field of laser precision machining technology. It includes a virtual-real state parameter acquisition module, a twin state modeling module, a chain-coupled response generation module, a deviation chain tracking module, a trajectory adaptive control module, and an optical performance reverse optimization module. The acquisition module obtains laser pulse parameters, residual phase noise, spot polarization state, material refractive index distribution, and surface temperature sequence. The modeling module generates a twin state space. The chain-coupled module constructs a chain response structure for light absorption, thermal diffusion, and phase transition reconstruction. The deviation chain module generates a processing deviation chain based on residual phase noise. The trajectory control module corrects the processing trajectory. The reverse optimization module adjusts processing parameters to form a virtual-real closed-loop control. This invention achieves quantitative prediction and dynamic compensation of processing deviations, improving the accuracy and consistency of lens processing.

Description

Technical Field

[0001] This application relates to the field of laser precision machining technology, and in particular to a digital twin system for laser lens processing. Background Technology

[0002] Traditional laser lens processing often employs open-loop control, meaning it executes the process according to preset parameters without real-time feedback or adjustment. However, laser processing involves strong coupling effects across multiple physical fields, including light absorption, heat conduction, and phase transition reconstruction, resulting in complex nonlinear interactions. Furthermore, factors such as residual phase noise from the laser source, ambient temperature fluctuations, and batch-to-batch material variations introduce random disturbances, causing deviations between the actual processed morphology and the design values.

[0003] Existing laser processing simulation technologies mainly employ single-physics modeling methods, such as solving only the heat conduction equation or approximating energy deposition solely through geometric optics. These methods neglect the bidirectional coupling effects between physical fields; for example, changes in refractive index caused by material phase transitions can, in turn, affect the subsequent energy deposition distribution. Furthermore, existing simulations are mostly performed offline, making it impossible to integrate with real-time acquired processing data. This leads to accumulated discrepancies between simulation results and physical reality, hindering the provision of reliable data for online parameter adjustments.

[0004] Residual phase noise of the laser source is a key factor affecting processing accuracy, especially in ultrafast laser processing. Phase noise causes instantaneous frequency fluctuations in the pulse, which in turn affects the coherence and spatial symmetry of energy deposition. Current technologies lack quantitative modeling methods for the transmission and amplification mechanisms of phase noise in the three stages of light absorption, thermal diffusion, and phase transition reconstruction, making it difficult to trace the causal relationship from source noise to processing deviations. Meanwhile, the matching relationship between the spot polarization state and the scanning trajectory direction significantly affects the processed morphology, but existing trajectory planning does not treat this as a dynamic compensation variable. Summary of the Invention

[0005] The embodiments of this application provide a digital twin system for laser lens processing, which enables quantitative prediction and dynamic correction of processing deviations, thereby improving the processing accuracy and consistency of complex surface lenses. To achieve the above objectives, this application adopts the following technical solution: A digital twin system for laser lens processing, the system comprising: The virtual and real state parameter acquisition module is used to acquire laser pulse parameters, laser source residual phase noise, spot polarization state, material refractive index distribution, and surface temperature sequence. The twin state modeling module is used to generate a twin state space based on the laser pulse parameters, the residual phase noise of the laser source, the polarization state of the light spot, the refractive index distribution of the material, and the surface temperature sequence. A chain-coupled response generation module is used to construct a photothermal phase change chain response structure composed of light absorption, thermal diffusion and phase change reconstruction based on the twin state space. The deviation chain tracking module is used to generate a processing deviation chain based on the residual phase noise of the laser source and the photothermal phase transition chain response structure. The trajectory adaptive control module is used to correct the machining trajectory based on the machining deviation chain to obtain the corrected machining trajectory. The optical performance reverse optimization module is used to adjust the processing parameters in reverse based on the corrected processing trajectory and the twin state space to form a virtual-real closed-loop control.

[0006] In some possible implementations, the twin state modeling module is specifically used for: The polarization state of the light spot and the refractive index distribution of the material are obtained, and the deposition distribution of laser energy inside the material is calculated. Based on the deposition distribution and the surface temperature sequence, determine the critical temperature threshold at which the optical absorption coefficient of the material undergoes an irreversible change; The spatial boundary corresponding to the critical temperature threshold is extracted as the inert region boundary parameter; The cumulative change in the refractive index distribution within the boundary parameters of the inert region relative to the initial refractive index distribution is calculated as the refractive index change rate of the inert region. The boundary parameters of the inert region and the rate of change of the refractive index of the inert region are combined to form the photoinduced inert region response parameters of the material.

[0007] In some possible implementations, the twin state modeling module is specifically used for: The refractive index gradient value along the laser scanning direction is calculated based on the refractive index distribution of the material and is used as the first component; the refractive index gradient value along the perpendicular scanning direction is calculated based on the refractive index distribution of the material and is used as the second component; the refractive index gradient value along the depth direction is calculated based on the refractive index distribution of the material and is used as the third component. Calculate the covariance between the first component and the second component as the first coupling term; calculate the covariance between the second component and the third component as the second coupling term; calculate the covariance between the first component and the third component as the third coupling term. The twin state modeling module combines the first component, the second component, the third component, the first coupling term, the second coupling term, and the third coupling term into a refractive index gradient drift tensor.

[0008] In some possible implementations, the twin state modeling module is specifically used for: Extract the temperature data of the heating phase and the temperature data of the cooling phase within a single laser pulse cycle from the surface temperature sequence; The relationship curve of the material's thermal expansion coefficient with temperature change is calculated based on the temperature data of the heating stage, and is used as the time-varying curve of the thermal expansion coefficient. The difference between the time required for the material to recover from its highest temperature to the ambient temperature and the laser pulse period is calculated based on the temperature data during the cooling phase, and is used as the shrinkage delay time. The time-varying curve of the thermal expansion coefficient is combined with the shrinkage delay time to form the time-varying thermodynamic response parameters of the material.

[0009] In some possible implementations, the chain-coupled response generation module is specifically used for: The polarization state of the light spot, the refractive index distribution of the material, and the time-varying thermodynamic response parameters of the material are obtained from the twin state space. The energy deposition distribution is calculated based on the polarization state of the light spot and the refractive index distribution of the material. The spatiotemporal evolution data of the temperature field are calculated based on the energy deposition distribution, the surface temperature sequence, and the time-varying thermodynamic response parameters of the material. The phase transition interface migration trajectory and reconstructed morphology data are calculated based on the spatiotemporal evolution data of the temperature field and the refractive index distribution of the material. The energy deposition distribution, the spatiotemporal evolution data of the temperature field, and the migration trajectory and reconstructed morphology data of the phase transition interface are combined into a photothermal phase transition chain response structure.

[0010] In some possible implementations, the chain-coupled response generation module is specifically used for: The material refractive index distribution is updated based on the reconstructed morphology data; The energy deposition distribution is recalculated based on the updated material refractive index distribution and the light spot polarization state; The spatiotemporal evolution data of the temperature field are recalculated based on the recalculated energy deposition distribution, the surface temperature sequence, and the material thermodynamic time-varying response parameters; The phase transition interface migration trajectory and morphology data are recalculated and reconstructed based on the recalculated spatiotemporal evolution data of the temperature field and the updated material refractive index distribution. Repeat the steps of updating the material refractive index distribution, recalculating the energy deposition distribution, recalculating the spatiotemporal evolution data of the temperature field, and recalculating the phase change interface migration trajectory and reconstructed morphology data until the root mean square error between two adjacent reconstructed morphology data is less than a preset threshold. Then, combine the final energy deposition distribution, spatiotemporal evolution data of the temperature field, and phase change interface migration trajectory and reconstructed morphology data into a photothermal phase change chain response structure.

[0011] In some possible implementations, the deviation chain tracking module is specifically used for: Extract the noise frequency distribution and noise amplitude sequence from the residual phase noise of the laser source; The transfer functions of the light absorption stage, the thermal diffusion stage, and the phase change reconstruction stage are calculated based on the aforementioned photothermal phase change chain response structure. The noise frequency distribution is convolved with the transfer function of the light absorption stage to obtain a first stage deviation sequence; the first stage deviation sequence is convolved with the transfer function of the thermal diffusion stage to obtain a second stage deviation sequence; the second stage deviation sequence is convolved with the transfer function of the phase transition reconstruction stage to obtain a processing deviation chain.

[0012] In some possible implementations, the trajectory adaptive control module is specifically used for: Obtain the spatial deviation component and phase deviation component in the machining deviation chain; Extract the major axis direction angle and ellipticity parameters of the polarization ellipse from the polarization state of the light spot; Obtain a preset reference scan trajectory; extract the tangent direction of each point from the reference scan trajectory; Calculate the angular deviation between the major axis direction angle and the tangent direction; The trajectory correction vector is calculated based on the included angle deviation value, the ellipticity parameter, the spatial deviation component, and the phase deviation component. The trajectory correction vector is superimposed on the position coordinates of each point on the reference scanning trajectory to obtain the corrected processing trajectory.

[0013] In some possible implementations, the optical performance reverse optimization module is specifically used for: The refractive index gradient drift tensor and the response parameters of the photoinert region of the material are obtained from the twin state space; Extract the first and second components from the main diagonal elements of the refractive index gradient drift tensor, and use them as the first gradient component and the second gradient component, respectively. Obtain a preset target refractive index distribution curve, calculate the deviation between the first gradient component and the target refractive index distribution curve as the first refractive index deviation, and calculate the deviation between the second gradient component and the target refractive index distribution curve as the second refractive index deviation. The depth values ​​along the laser scanning direction, along the perpendicular scanning direction, and along the depth direction in the inert region boundary parameters are extracted from the photoinduced inert region response parameters of the material and used as the first depth value, the second depth value, and the third depth value, respectively. Extract the rate of change of refractive index of the inert region corresponding to the boundary parameters of the inert region, and use it as the deviation of the rate of change of refractive index; The adjustment amounts of the laser pulse energy and the scanning speed are calculated in reverse based on the first refractive index deviation, the second refractive index deviation, the refractive index change rate deviation, the first depth value, the second depth value, and the third depth value, and are output as the processing parameters after reverse adjustment.

[0014] In some possible implementations, the twin state modeling module is also used for: Obtain the reverse-adjusted processing parameters output by the optical performance reverse optimization module; The reverse-adjusted processing parameters are input into the twin state space; The material's photo-induced inert region response parameters, refractive index gradient drift tensor, and thermodynamic time-varying response parameters are recalculated based on the reverse-adjusted processing parameters. The recalculated material refractive index distribution, the material photoinduced inert region response parameters, the refractive index gradient drift tensor, and the material thermodynamic time-varying response parameters are updated to the twin state space and output to the chain-coupled response generation module. The photothermal phase transition chain response structure is reconstructed based on the updated twin state space.

[0015] As can be seen from the above technical solution, this application has the following beneficial effects: 1. This system constructs a twin state space, transforming heterogeneous physical quantities such as laser pulse parameters, residual phase noise, beam polarization state, material refractive index distribution, and surface temperature sequence into characteristic parameters with clear physical meaning, providing a unified digital foundation for subsequent coupled analysis. Based on this, a photothermal phase transition chain response structure composed of light absorption, thermal diffusion, and phase transition reconstruction is constructed, explicitly linking three causally connected and mutually feedback-driven physical processes. Iterative solutions achieve self-consistent calculations of refractive index changes and energy deposition. This structured twin modeling method enables the digital twin to reflect the dynamic evolution of the physical processing process with high fidelity, solving the technical problems of insufficient accuracy in traditional single-physics simulations and their inability to handle the strong coupling effect of photothermal phase transitions, thus improving the accuracy of processing morphology prediction.

[0016] 2. This system, through a deviation chain tracking module, performs temporal convolution of the residual phase noise of the laser source with the impulse response functions of the three stages: light absorption, thermal diffusion, and phase transition reconstruction. This quantitatively establishes the complete transmission path from the source noise to the final processing deviation, making the phase noise, which was originally considered a random error, predictable and traceable. Based on this, the trajectory adaptive control module dynamically corrects the scanning trajectory using the processing deviation chain and the polarization state of the light spot, achieving joint compensation for polarization anisotropy deviation and noise-induced deviation. The optical performance reverse optimization module further adjusts the processing parameters in reverse based on the refractive index gradient drift tensor and inert region parameters in the twin state space, forming a closed-loop control. This allows for online compensation for process drift caused by material batch differences, environmental disturbances, and equipment aging, effectively ensuring the processing accuracy and optical performance consistency of complex surface lenses. Attached Figure Description

[0017] The invention will now be further described with reference to the accompanying drawings.

[0018] Figure 1 The overall closed-loop control flowchart of the system provided in the embodiments of this application; Figure 2 A flowchart of twin state modeling provided for embodiments of this application; Figure 3 This is a flowchart of chain-coupled response generation and iteration provided in an embodiment of this application; Figure 4 A flowchart illustrating deviation tracking and trajectory correction provided in an embodiment of this application. Detailed Implementation

[0019] The terms "first," "second," and "third," etc., used in this application specification, claims, and drawings are for distinguishing different objects, not for specifying a particular order.

[0020] In the embodiments of this application, the words "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the words "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0021] Research has revealed that existing laser lens processing technologies employ open-loop control, which cannot detect and compensate for random disturbances such as residual phase noise of the laser source and environmental fluctuations in real time. Furthermore, single-physics simulations neglect the bidirectional coupling effect between photothermal phase transitions, resulting in insufficient accuracy in predicting the processed morphology. In addition, offline simulations cannot be dynamically integrated with real-time processing data, lacking quantitative modeling of the transmission mechanism of phase noise in light absorption, thermal diffusion, and phase transition reconstruction. It is also difficult to use the polarization state of the light spot as a dynamic compensation variable, making it impossible to trace the causal relationship from the source noise to the final deviation.

[0022] To address the aforementioned issues, this application provides a digital twin system for laser lens processing: Example 1 To solve the above problems, such as Figures 1-4 As shown, this embodiment describes a digital twin system for laser lens processing, which is used to achieve high-fidelity simulation, deviation tracking and adaptive control of the processing state during laser-induced back wet etching of customized aspherical microlens arrays.

[0023] I. Technical Fields and Terminology Definitions: The original dataset consists of a collection of physical quantities directly acquired by sensors without feature extraction. This includes laser pulse parameters (pulse energy, repetition frequency, pulse width), residual phase noise of the laser source (time-domain waveform), beam polarization state (Stokes parameters), material refractive index distribution (real part of refractive index on a 3D mesh), and surface temperature sequence (time-temperature point set). This dataset is used only for immediate computation and is not stored permanently in the twin state space.

[0024] Twin state space: A structured database that stores feature parameters extracted from the raw data that have long-term predictive value, specifically including four fields: Current processing parameters (pulse energy, scanning speed, focal position); Response parameters of the photo-induced inert region of the material; Refractive index gradient drift tensor; Material thermodynamic time-varying response parameters; The updated three-dimensional array of material refractive index distribution is dynamically updated during processing. The twin state space does not include instantaneous raw data acquisition to avoid overlapping responsibilities.

[0025] Photothermal phase change chain response structure: a composite data object consisting of three members: energy deposition distribution (a three-dimensional array, each element representing the heat generation power per unit volume per unit time), temperature field spatiotemporal evolution data (a four-dimensional array: three-dimensional space plus one-dimensional time), and phase change interface migration trajectory and reconstructed morphology data (time series of interface positions and two-dimensional height map of the final surface).

[0026] Machining deviation chain: A one-dimensional array arranged in pulse time order, where each element is a three-dimensional deviation vector, and the components represent the predicted positional offset of the machining point in the scanning direction, vertical scanning direction, and depth direction, respectively.

[0027] Inert region: A region where the refractive index undergoes an irreversible and permanent change under laser irradiation. The criterion for irreversibility is: after reheating the region to the same temperature and then cooling it, the change in refractive index compared to the original value should be less than a preset threshold, such as one percent. This definition is only used in offline calibration; in online operation, it is approximated by a temperature threshold.

[0028] Inert region boundary mesh: A closed triangular mesh representing the boundary of the inert region. Each triangle is defined by the coordinates of three vertices, and the mesh as a whole encloses a three-dimensional region. When calculating the depth value, the axial bounding box of the mesh along the three coordinate axes is taken, which is the difference between the maximum and minimum values ​​of the vertex coordinates. For concave boundaries, the axial bounding box will include the volume of the non-inert region; in this case, the depth value is an approximate estimate, but acceptable for control purposes.

[0029] II. Overall System Data Flow and Module Timing: The virtual and real state parameter acquisition module sends the original acquired dataset to the twin state modeling module and the chain-coupled response generation module, respectively.

[0030] The twin state modeling module extracts feature parameters from the original data and stores them in the twin state space.

[0031] The chain-coupled response generation module reads feature parameters from the twin state space and uses the surface temperature sequence from the original acquired dataset as boundary conditions to generate a photothermal phase change chain response structure.

[0032] The deviation chain tracking module reads the residual phase noise of the laser source (from the original acquisition dataset) and the photothermal phase transition chain response structure, and outputs the processing deviation chain.

[0033] The trajectory adaptive control module reads the machining deviation chain, the spot polarization state (from the original acquisition dataset), and the preset reference scanning trajectory, and outputs the corrected machining trajectory.

[0034] The optical performance reverse optimization module reads the twin state space and the corrected machining trajectory, and reverse-calculates the new machining parameters.

[0035] The twin state modeling module receives new processing parameters, combines them with the latest data fed back from the physical space in real time, updates the twin state space, and triggers the chain-coupled response generation module to recalculate.

[0036] III. Virtual and Real State Parameter Acquisition Module: This module synchronously collects key parameters during the physical processing using the following sensors, and all data is accompanied by a unified timestamp: Laser pulse parameters: A high-speed photodetector is installed on the laser optical path to measure the actual energy value, pulse arrival time and pulse width of each laser pulse.

[0037] Residual phase noise of the laser source: A coherent probe optical path is integrated at the laser output end, and the laser output is mixed with an ultra-stable local oscillator to extract the time-domain waveform of the residual phase noise. The sampling rate is not less than ten times the pulse repetition frequency, and it is recorded as a sequence of voltage changes over time, which is then calibrated and converted into phase radian values.

[0038] Beam polarization state: A polarization measuring instrument is set before the beam enters the processing head to output Stokes parameters. Then, the major axis direction angle, ellipticity (minor axis length divided by major axis length, 1 for circular polarization, 0 for linear polarization) and rotation direction of the polarization ellipse are calculated.

[0039] Material refractive index distribution: Before processing begins, the lens substrate is scanned in three dimensions using optical coherence tomography (OCT) to obtain three-dimensional mesh data of the real part of the refractive index. The mesh spacing is less than one-fifth of the laser wavelength; for example, for a laser with a wavelength of 1 micrometer, the mesh spacing is no greater than 0.2 micrometers. The scanning results are stored as a three-dimensional array, indexed to corresponding spatial coordinates.

[0040] Surface temperature sequence: An infrared thermal imager is installed next to the processing area to acquire temperature data of the processed surface at a frame rate synchronized with the laser pulses. Each pixel records a curve of temperature change over time, with a sampling frequency no less than twice the pulse repetition frequency.

[0041] Furthermore, to support subsequent energy deposition calculations, this module also requires prior knowledge of the temperature dependence of the material's extinction coefficient. During system initialization, a spectrophotometer is used to measure the material's transmission spectrum at different temperatures (from room temperature to above the material's melting point). The extinction coefficient as a function of temperature and wavelength is then derived using Beer-Lambert's law. The data is stored as a two-dimensional lookup table. During online calculations, the extinction coefficient of each grid point is obtained by interpolation from the table based on the temperature value of that point in the current temperature field.

[0042] IV. Twin State Modeling Module: This module transforms the raw acquired parameters into a structured twin state space. Its implementation consists of four sub-functions.

[0043] (a) Extraction of response parameters in the photo-induced inert region of the material: Definition: This parameter includes two fields: the boundary grid of the inert region (triangular grid) and the average rate of change of refractive index of the inert region (dimensionless real number).

[0044] Extraction steps: Calculating the energy deposition distribution: Based on the beam polarization state and the material refractive index distribution, Maxwell's equations are solved using the finite-difference time-domain method. The processing region is discretized into a grid with the same spacing as the refractive index distribution. Each grid point is assigned a complex refractive index, with the real part derived from the refractive index distribution and the imaginary part from an extinction coefficient lookup table. (Initial temperature is taken as ambient temperature). Using the measured polarization state of the light spot as the incident boundary condition, the electric and magnetic fields are iteratively updated until stability is achieved. Then, the Joule thermal power density at each grid point is calculated: in The laser angular frequency, The vacuum permittivity, Let (E) be the extinction coefficient at that point, and (|E|) be the modulus of the electric field intensity. The three-dimensional energy deposition distribution is then obtained.

[0045] Determining the critical temperature threshold: Extract the temperature peak within each pulse cycle from the surface temperature sequence, and simultaneously deduce the material's optical absorption coefficient at that moment by analyzing the change in surface reflectivity. ,in Let the wavelength be λ. Plot the absorption coefficient against the temperature peak to obtain a curve. Calculate the slope of the curve at adjacent points, dividing the change in absorption coefficient by the change in temperature. When the slope exceeds the initial slope, i.e., five times that near room temperature, an irreversible change is considered to have occurred, and the corresponding temperature is the critical temperature threshold. Once this threshold is determined, it is used as a fixed parameter in subsequent processing.

[0046] Extracting the inert region boundary: In the three-dimensional temperature field solved by the heat conduction equation, identify all grid points whose temperature values ​​are exactly equal to the critical temperature threshold. Connect these points using the moving cube algorithm to form a closed triangular mesh, which is the inert region boundary mesh.

[0047] Calculate the average refractive index change rate in the inert region: Traverse every voxel within the boundary grid of the inert region. Determine if a voxel is inside the grid using ray casting. Read the current refractive index value of that voxel and subtract its initial refractive index value before processing to obtain the refractive index change. Sum the changes of all voxels and divide by the total volume of the inert region (enclosed by the boundary grid). Calculate the average refractive index change rate of the inert region using the divergence theorem.

[0048] Combined response parameters: The boundary mesh of the inert region and the average refractive index change rate of the inert region are packaged into the material's photoinduced inert region response parameters and stored in the twin state space.

[0049] (II) Construction of the refractive index gradient drift tensor: Definition: This tensor is a 3x3 symmetric matrix used to describe the spatial rate of change of the material's refractive index in three orthogonal directions and the degree of their coupling. The matrix form is: in These represent the spatial gradient fields of refractive index in the (X, Y, Z) directions. Represents variance. This represents the covariance.

[0050] Build steps: Calculate the refractive index gradient fields in three directions: Read the current three-dimensional array of the material's refractive index distribution from the twin state space. For each grid point ((i,j,k)), calculate the gradient using the central difference formula: in Set the grid spacing. Apply this to all grid points. The values ​​are collected into a two-dimensional array (actually a one-dimensional expansion), which is called the first component; similarly, the second and third components are obtained.

[0051] Calculate variance and covariance: For all grid points in the entire processing area, with a total of (N), calculate: Similarly, calculate Due to the symmetry of covariance, Therefore, the matrix is ​​symmetric.

[0052] Combining into a tensor: Fill the corresponding positions of the matrix with the above six values ​​to obtain the refractive index gradient drift tensor, which is then stored in the twin state space. The drift in this tensor refers to the deviation of the actual gradient distribution from the ideal design distribution, which is reflected in the off-diagonal elements being non-zero, indicating that the gradient direction has rotated.

[0053] (III) Extraction of time-varying thermodynamic response parameters of materials: Definition: This parameter includes two fields: the coefficient of thermal expansion as a function of temperature and the shrinkage delay time.

[0054] Extraction steps: Separate the heating and cooling phases: Extract data from a single laser pulse cycle from the surface temperature sequence. The segment within that cycle where the temperature rises from the initial value to the peak value is marked as the heating phase temperature data, and the segment where the peak value falls back to the initial value is marked as the cooling phase temperature data.

[0055] Extracting the thermal expansion coefficient curve: During the heating stage, the thermal strain (relative change of displacement at various points on the surface) of the processed surface is simultaneously measured using digital image correlation. Establishing the thermoelastic constitutive equation: in To measure strain, Let M be the coefficient of thermal expansion function to be determined. Discretize the temperature interval into (M) smaller intervals, and assume... If is a constant, then the integral becomes a summation. This is obtained by solving a system of linear equations with smoothing constraints (using Tikhonov regularization). The value for each temperature range. The regularization objective function is: Where (A) is the integral matrix, and (D) is the first-order difference matrix. The regularization parameter is selected using the L-curve method. The final result is a piecewise linear curve, i.e., a time-varying curve of the thermal expansion coefficient, which is stored in the form of a temperature coefficient correspondence table.

[0056] Calculate the shrinkage delay time: During the cooling phase, record the total time required for the surface temperature to drop from its peak to ambient temperature. .Will Subtract the period length of a laser pulse The difference obtained This is the contraction delay time. If This indicates that the area heated by the previous pulse has not completely cooled and contracted before the next pulse arrives; if This indicates that cooling has been fully completed.

[0057] Combined response parameters: The thermal expansion coefficient curve, i.e. the temperature coefficient correspondence table and the shrinkage delay time are packaged into material thermodynamic time-varying response parameters and stored in the twin state space.

[0058] (iv) Acquisition and storage of thermophysical parameters: To support the calculation of the thermal diffusion process, the system also needs to obtain the material's thermal conductivity (k(T)) and specific heat capacity in advance. and density A function that varies with temperature. During system initialization, it is obtained through the following standard measurement methods: Specific heat capacity: Measured using differential scanning calorimetry (DSC). curve.

[0059] Thermal diffusivity: The thermal diffusivity is obtained by measuring using the laser flare method (LFA). .

[0060] Thermal conductivity: derived from the relationship Calculated.

[0061] Density: Obtained by integrating the coefficient of thermal expansion curve. ,in This is the density at room temperature.

[0062] The functions mentioned above are all stored as lookup tables for use in online heat conduction calculations based on temperature interpolation.

[0063] V. Chain-coupled response generation module: This module iteratively solves the photothermal phase transition coupling process based on the characteristic parameters in the twin state space and some original acquired data.

[0064] (a) Basic chain response generation: enter: Obtain from the twin state space: the polarization state of the light spot, the current refractive index distribution of the material, and the time-varying thermodynamic response parameters of the material (thermal expansion coefficient curve and shrinkage delay time).

[0065] Obtain the surface temperature sequence as thermal boundary condition from the original acquired dataset.

[0066] Obtain the following from the initialization data: thermal conductivity (k(T)) and specific heat capacity. ,density Lookup table and interface migration coefficient (Typical range of values ​​obtained in advance through isothermal phase transition kinetic experiments) ).

[0067] Execution steps: Light absorption stage: Using the same finite-difference time-domain method as step 1 in Part 4 (I), the three-dimensional energy deposition distribution (Q(x,y,z)) is calculated based on the polarization state of the light spot and the current refractive index distribution of the material. Note: The refractive index distribution here may be a value updated after multiple iterations.

[0068] Thermal diffusion process: The energy deposition distribution (Q(x,y,z,t)) is treated as a heat source term, which is non-zero during the pulse duration and zero during the pulse interval. This is substituted into the three-dimensional unsteady-state heat conduction equation: The finite volume method is used for discretization, and an implicit Euler scheme is employed for time progression to ensure stability. Boundary conditions: the upper surface (machined surface) is given a convective heat transfer boundary, and the surface temperature at each time step is obtained by interpolation of the surface temperature sequence; the remaining surfaces are set to adiabatic or isothermal (depending on the experimental setup). At each time step, the grid point is obtained by interpolation based on the current temperature field. Then, the system of linear equations is solved to obtain the temperature field at the next time step. The calculation results for all time steps are stored in a four-dimensional array. That is, the spatiotemporal evolution data of the temperature field.

[0069] Phase transition reconstruction stage: Based on the spatiotemporal evolution data of the temperature field, the level set method is used to track the migration of the phase transition interface. A level set function is defined. Its zero isosurface represents the solid-liquid or liquid-gas interface. The migration velocity at the interface is given by the phase transition kinetic equation: Where (T) is the temperature at the interface, The melting point or glass transition temperature of the material. Let be the interface migration coefficient. The evolutionary form of the level set equation is: An upwind approach is used for the solution. Initial conditions: The sign distance function is taken as positive in the initial solid phase region and negative in the gas phase region. At each time step, the superheat at the interface is calculated based on the current temperature field, the level set function is updated, and the distance field is reinitialized. Finally, the zero isosurface position at each time step is recorded to obtain the phase transition interface migration trajectory; after processing, [the following is taken]... The zero isosurface is used as the final reconstructed topographic data, and it is projected onto a two-dimensional plane to obtain a height map.

[0070] Combined output: The energy deposition distribution, temperature field spatiotemporal evolution data, phase change interface migration trajectory and reconstructed morphology data are packaged into a photothermal phase change chain response structure.

[0071] (ii) Iterative solution considering optical feedback: Since material phase transitions can change their refractive index, such as densification which increases the refractive index, thus affecting subsequent laser energy deposition, iterative processes must be performed until self-consistent.

[0072] Iteration steps: Perform basic chain response generation (Section 5.1) to obtain the first predicted reconstructed morphology data.

[0073] Updated refractive index distribution: Based on the reconstructed morphology data, regions where permanent phase transitions occurred were identified, i.e., areas where the temperature had exceeded the critical temperature threshold and had cooled and solidified. Within these regions, the refractive index was updated according to known material relationships. For fused silica, the refractive index change caused by densification is approximately proportional to the density change: in This is a proportionality constant (approximately 0.1~0.2). Density variation. The result is obtained by integrating the thermal expansion curve: The integral is applied from room temperature to peak temperature. For regions where no permanent phase transition occurs, only reversible thermal expansion is performed, and the refractive index reverts to its original value. The updated refractive index distribution replaces the old value in the twin state space.

[0074] Recalculate energy deposition: Using the updated refractive index distribution, perform the calculation of the light absorption stage again to obtain a new energy deposition distribution.

[0075] Recalculate the temperature field: Using the new energy deposition distribution as the heat source, solve the heat conduction equation again to obtain new spatiotemporal evolution data of the temperature field.

[0076] Recalculate the phase transition: Based on the new temperature field and the updated refractive index distribution, perform the phase transition reconstruction step again to obtain new reconstructed morphology data.

[0077] Convergence criterion: Compare the reconstructed topography data obtained in this calculation with the result obtained in the previous calculation. Both topography data are two-dimensional height maps with dimensions of [missing information]. Subtract the height values ​​at the same plane location to obtain the difference array. Calculate the root mean square error: Preset a tiny threshold For example, one-tenth of the allowable topographic tolerance in lens manufacturing, such as 10 nanometers. If... If the iteration is successful, the iteration is considered to have converged; otherwise, return to step 2 and continue iterating.

[0078] Final output: The energy deposition distribution, temperature field spatiotemporal evolution data, phase change interface migration trajectory and reconstructed morphology data obtained from the last iteration are packaged into the final photothermal phase change chain response structure.

[0079] VI. Deviation Chain Tracking Module: This module is responsible for converting the residual phase noise of the laser source into a quantitative prediction of processing deviations. Since changes in the material state during processing may alter the transmission characteristics of each stage, this module adopts a piecewise time-invariant assumption: the processing is divided into several windows over time, for example, each window consists of 100 pulses. Within each window, the system is considered time-invariant, and the impulse response function is re-identified when switching windows.

[0080] Definition: The machining deviation chain is a one-dimensional array indexed by pulse number, where each element is a three-dimensional vector. , respectively, represent the positional offset of the predicted processing point in the scanning direction, vertical scanning direction, and depth direction.

[0081] Execution steps: Preprocessing phase noise: Obtain the time-domain waveform of the residual phase noise of the laser source from the virtual and real state parameter acquisition module. The waveform is arranged according to the pulse period. The equivalent phase perturbation value for each pulse is obtained by segmenting the pulse and averaging it within each period. .in This is the pulse sequence number.

[0082] Identify the impulse response function of each stage, either online or offline: At the beginning of each time window, using the material state corresponding to the current window, the current refractive index distribution, temperature field, etc. are obtained from the twin state space, and the following identification is performed in the high-fidelity simulation environment: For the optical absorption stage: A unit pulse perturbation is applied to the laser phase input of the simulation model, that is, the phase is instantaneously increased by 1 radian at the 0th pulse, and 0 for the remaining pulses. The change of the spatial centroid shift of the energy deposition distribution caused by this perturbation over time is recorded. The time series of the shift is normalized with the pulse period as the sampling interval to obtain the pulse response sequence of the optical absorption stage. , .

[0083] For the thermal diffusion stage: the above energy deposition shift sequence is used as input to the thermal diffusion simulation model, and the time response of the temperature field centroid shift is recorded. After normalization, the impulse response sequence of the thermal diffusion stage is obtained. .

[0084] For the phase transition reconstruction stage: the temperature field centroid shift sequence is used as input and applied to the phase transition reconstruction simulation model. The time response of the final morphological height shift is recorded, and after normalization, the impulse response sequence of the phase transition reconstruction stage is obtained. .

[0085] Once the identification is complete, these three sets of impulse response functions will be used consistently within this window.

[0086] Temporal convolution chain computation: For the The first stage deviation (energy deposition center of gravity shift) generated by each pulse is: in From 0 to Summation. In actual calculations, only a finite length is used, such as the length at which the impulse response function decays to 1% of its peak value.

[0087] Second stage deviation (temperature field centroid shift): Final processing deviation chain: in It is a three-dimensional vector.

[0088] Output: Processing deviation chain The data is passed to the trajectory adaptive control module in chronological order. When switching windows, the deviation calculation for the first few pulses of the new window requires the simultaneous use of the pulse response function (historical data) of the old window and the function (current data) of the new window, employing a weighted transition.

[0089] VII. Trajectory Adaptive Control Module: This module corrects the preset scanning trajectory in real time based on the processing deviation chain and the current spot polarization state.

[0090] Definition: A preset baseline scan trajectory is an ordered set of spatial points. Each point includes three-dimensional coordinates. The corrected processing trajectory is a new set of points obtained by superimposing a correction vector on each reference point.

[0091] Execution steps: Extracting the deviation component: Extracting the pulse to be processed (the first pulse) from the machining deviation chain. The deviation vector corresponding to each pulse .in The predicted offset along the scanning direction, This represents the predicted offset in the vertical scanning direction. This represents the predicted offset in the depth direction.

[0092] Obtaining the polarization characteristics of the light spot: The polarization state of the current light spot is obtained from the virtual and real state parameter acquisition module, and the major axis direction angle of the polarization ellipse is extracted. (Unit: radians) and ellipticity parameters Circular polarization When linearly polarized ).

[0093] Calculate the tangent direction of the trajectory: locate the expected position of the current pulse on the reference scan trajectory. Take the point preceding this point. and the next point (Use one-sided difference for boundary points), calculate the direction vector. After normalization, the unit tangent direction is obtained. Simultaneously calculate the normal direction (perpendicular to the tangent and located in the horizontal plane) and the depth direction (vertically upward) at that point.

[0094] Calculate the included angle deviation: Calculate the minimum included angle between the major axis direction angle of the polarization ellipse and the tangent direction. Take the acute angle ((0) to) (between). It is the azimuth angle of the tangent direction in the horizontal plane.

[0095] Calculate the trajectory correction vector: The three components of the correction vector are calculated using the following formula: in This is a proportionality coefficient, obtained through offline calibration experiments. Calibration method: Under standard machining conditions, known deviations are applied, for example, a known offset is artificially superimposed on the trajectory. The actual required correction amount is measured to achieve the desired machining position. The coefficient value is obtained by least squares fitting. Typical value range: , (Unit: meters per radian) , Physical meaning: When the laser is linearly polarized ( When the polarization direction is not parallel to the scanning direction, an additional lateral offset will occur, which is achieved through... Item compensation; depth deviation is compensated through partial coupling in the vertical direction ( Item) and direct depth compensation ( item).

[0096] Superposition correction: The correction vector is superimposed. Add to the baseline trajectory point On the coordinates, the corrected machining trajectory points are obtained: For all Perform the above operations to obtain the complete corrected processing trajectory.

[0097] Output: The corrected machining trajectory is sent to the motion control system for execution.

[0098] VIII. Optical Performance Reverse Optimization Module: Based on the optical quality characteristics in the current twin state space and the executed correction trajectory, this module reverse-optimizes subsequent processing parameters (laser pulse energy and scanning speed) to make the optical performance of the final lens approach the design target.

[0099] Definition: The target refractive index distribution curve refers to the ideal refractive index variation curve along the radial direction of the lens, from the center to the edge. It is derived from the optical design equations of lenses, such as the aspherical formula combined with the material dispersion relation. Its derivative (tangent slope) represents the focusing power distribution of the lens.

[0100] Execution steps: Acquiring twin state characteristics: The refractive index gradient drift tensor and the material photoinert region response parameters, as well as the partially corrected processing trajectory, are read from the twin state space to evaluate the deviation between the actual scanning path and the design.

[0101] Extracting the gradient component: Extract the first element on the main diagonal from the refractive index gradient drift tensor. and the second element Note: Variance is used here instead of standard deviation because variance is directly related to energy. However, it needs to be converted to gradient values ​​for comparison with the target value. Standard deviation is actually used in practice. As the first gradient component As the second gradient component. These two values ​​represent respectively direction and Typical magnitude of the refractive index gradient in the direction.

[0102] Calculate the refractive index gradient deviation: Target refractive index distribution curve radial derivative The value near the center of the lens is approximately constant, denoted as . (For rotationally symmetric lenses, (Independent of direction at the center). The first gradient component is actually measured. and Compare: Similarly .like and A positive value indicates that the actual focusing ability is too strong; a negative value indicates that it is too weak; the difference between the two indicates astigmatism.

[0103] Extracting inert region parameters: Extract the boundary mesh of the inert region from the photoinduced inert region response parameters of the material. Calculate the maximum span of this mesh along the laser scanning direction. Maximum span in the vertical scanning direction Maximum span in the depth direction Simultaneously, the average refractive index change rate in the inert region is extracted. (Dimensionless). Calculate the deviation of the rate of change of refractive index: in The average refractive index change required for lens design (obtained by dividing the required phase modulation by the optical path length).

[0104] Evaluate scan consistency by combining the corrected machining trajectory: Extract the curvature changes and speed fluctuations of the actual scan path from the corrected machining trajectory. If the deviation between the actual scan speed and the preset speed exceeds a threshold, such as 5%, it indicates that the trajectory correction has introduced additional dynamic errors, which need to be considered in reverse optimization. Define a scan consistency factor: (S) The closer to 1, the better the consistency. This factor is used to adjust the optimization intensity.

[0105] Backpropagation calculation: A pre-trained backpropagation neural network is used to map the bias to the processing parameters. Network structure: Input layer: 5 nodes ( ); Output layer: 2 nodes (pulse energy adjustment ratio) Scan speed adjustment ratio ); Hidden layers: 3 layers, 12 nodes per layer, with ReLU activation function; Training method: For the current lens model (with a specific target refractive index distribution curve), a large number of different combinations of pulse energy and scanning speed are randomly generated in the simulation environment (energy within the rated value). Uniform sampling within the range, with the speed at the rated value. (Uniform sampling within the inner quadrilateral), run the twin model (Section 5) to calculate the corresponding And consistency with actual scanning. A training dataset is created to minimize... With the goal of convergence, the Adam optimizer is trained.

[0106] Online inference: Input the currently calculated deviation and consistency factor into the network, and the network outputs... and .For example, This indicates a 5% increase in pulse energy. This indicates a 5% reduction in scanning speed.

[0107] Output: Adjusted pulse energy and scanning speed The reverse-adjusted processing parameters are output to the motion controller and laser control interface for subsequent processing.

[0108] IX. Dynamic Updates and Closed-Loop Iteration of the Twin State Space: To achieve continuous adaptive control, the twin state space needs to be updated based on the new parameters after inverse optimization, and the chain-coupled response generation module needs to be redriven.

[0109] Execution steps: Receiving new processing parameters: The twin state modeling module obtains the reverse-adjusted processing parameters (new pulse energy) from the optical performance reverse optimization module. and new scanning speed ).

[0110] Update the current processing parameter field in the twin state space: and Write to the twin state space, overwriting the old value.

[0111] Feature parameters are re-extracted: Based on the new processing parameters and combined with the latest data from real-time feedback in physical space, such as new surface temperature sequences and new refractive index distribution measurements, the twin state modeling module re-executes all the calculation processes described in Part 4. The response parameters of the photoinduced inert region of the material are re-extracted because changes in energy and velocity affect the boundary of the inert region and the rate of change of refractive index. Reconstruct the refractive index gradient drift tensor because the refractive index distribution will change with the new parameters; The thermodynamic time-varying response parameters of the material are re-extracted, as the thermal expansion curve and contraction delay time may change due to different temperature histories; Update the twin state space: Overwrite the old values ​​in the twin state space with the newly calculated parameters, and update the three-dimensional array of the material refractive index distribution.

[0112] Triggering chain-coupled response regeneration: Using the updated twin state space as input, the chain-coupled response generation module is invoked to reconstruct the photothermal phase transition chain response structure using an iterative method. This structure will reflect the predicted processing results under the new processing parameters.

[0113] Cyclic control: The above process occurs in each processing batch, for example, after each scan line is completed or each Execute once after each pulse. The system sets a maximum number of iterations, such as 10, and a performance tolerance threshold (e.g., ...). and The absolute values ​​are all less than (0.001) per micrometer, and The absolute value is less than (0.01). When the number of iterations reaches the upper limit or the performance deviation is less than the threshold, the optimization stops, the final processing parameters are output and fixed for the remaining processing.

[0114] 10. Summary of the overall system workflow: In a complete lens manufacturing task, the above modules work together in the following sequence: Startup and initialization: Load the preset baseline scan trajectory and initial machining parameters. Load the pre-measured material properties (extinction coefficient table, thermal conductivity table, specific heat capacity table, density table, interface migration coefficient). The virtual and real state parameter acquisition module begins real-time data acquisition.

[0115] First twin modeling: The twin state modeling module generates an initial twin state space based on initial parameters and the first frame of acquired data.

[0116] Prediction and Compensation Cycle: The chain-coupled response generation module generates a photothermal phase change chain response structure.

[0117] The deviation chain tracking module calculates the machining deviation chain using the impulse response function of the current window.

[0118] The trajectory adaptive control module outputs the corrected machining trajectory.

[0119] The motion system processes a pulse or a short path according to the corrected trajectory.

[0120] Performance evaluation and reverse optimization: After completing a scan area, such as a scan line, the optical performance reverse optimization module calculates new processing parameters based on the current twin state space and the already executed correction trajectory. .

[0121] State Update: The twin state modeling module updates the twin state space based on the new parameters and triggers the chain-coupled response generation module to recalculate.

[0122] Iteration: Return to step 3 and continue processing using the updated twin state space and new parameters until the entire lens is processed.

[0123] The foregoing has shown and described the basic principles, main features, and advantages of this application. Those skilled in the art should understand that this application is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of this application. Various changes and modifications can be made to this application without departing from the spirit and scope thereof, and all such changes and modifications fall within the scope of this application as claimed. The scope of protection of this application is defined by the appended claims and their equivalents.

Claims

1. A digital twin system for laser lens processing, characterized in that, The system includes: The virtual and real state parameter acquisition module is used to acquire laser pulse parameters, laser source residual phase noise, spot polarization state, material refractive index distribution, and surface temperature sequence. The twin state modeling module is used to generate a twin state space based on the laser pulse parameters, the residual phase noise of the laser source, the polarization state of the light spot, the refractive index distribution of the material, and the surface temperature sequence. A chain-coupled response generation module is used to construct a photothermal phase change chain response structure composed of light absorption, thermal diffusion and phase change reconstruction based on the twin state space. The deviation chain tracking module is used to generate a processing deviation chain based on the residual phase noise of the laser source and the photothermal phase transition chain response structure. The trajectory adaptive control module is used to correct the machining trajectory based on the machining deviation chain to obtain the corrected machining trajectory. The optical performance reverse optimization module is used to adjust the processing parameters in reverse based on the corrected processing trajectory and the twin state space to form a virtual-real closed-loop control.

2. The system according to claim 1, characterized in that, The twin state modeling module is specifically used for: The polarization state of the light spot and the refractive index distribution of the material are obtained, and the deposition distribution of laser energy inside the material is calculated. Based on the deposition distribution and the surface temperature sequence, determine the critical temperature threshold at which the optical absorption coefficient of the material undergoes an irreversible change; The spatial boundary corresponding to the critical temperature threshold is extracted as the inert region boundary parameter; The cumulative change in the refractive index distribution within the boundary parameters of the inert region relative to the initial refractive index distribution is calculated as the refractive index change rate of the inert region. The boundary parameters of the inert region and the rate of change of the refractive index of the inert region are combined to form the photoinduced inert region response parameters of the material.

3. The system according to claim 1, characterized in that, The twin state modeling module is specifically used for: The refractive index gradient value along the laser scanning direction is calculated based on the refractive index distribution of the material and is used as the first component; the refractive index gradient value along the perpendicular scanning direction is calculated based on the refractive index distribution of the material and is used as the second component; the refractive index gradient value along the depth direction is calculated based on the refractive index distribution of the material and is used as the third component. Calculate the covariance between the first component and the second component as the first coupling term; calculate the covariance between the second component and the third component as the second coupling term; calculate the covariance between the first component and the third component as the third coupling term. The twin state modeling module combines the first component, the second component, the third component, the first coupling term, the second coupling term, and the third coupling term into a refractive index gradient drift tensor.

4. The system according to claim 1, characterized in that, The twin state modeling module is specifically used for: Extract the temperature data of the heating phase and the temperature data of the cooling phase within a single laser pulse cycle from the surface temperature sequence; The relationship curve of the material's thermal expansion coefficient with temperature change is calculated based on the temperature data of the heating stage, and is used as the time-varying curve of the thermal expansion coefficient. The difference between the time required for the material to recover from its highest temperature to the ambient temperature and the laser pulse period is calculated based on the temperature data during the cooling phase, and is used as the shrinkage delay time. The time-varying curve of the thermal expansion coefficient is combined with the shrinkage delay time to form the time-varying thermodynamic response parameters of the material.

5. The system according to claim 1, characterized in that, The chain-coupled response generation module is specifically used for: The polarization state of the light spot, the refractive index distribution of the material, and the time-varying thermodynamic response parameters of the material are obtained from the twin state space. The energy deposition distribution is calculated based on the polarization state of the light spot and the refractive index distribution of the material. The spatiotemporal evolution data of the temperature field are calculated based on the energy deposition distribution, the surface temperature sequence, and the time-varying thermodynamic response parameters of the material. The phase transition interface migration trajectory and reconstructed morphology data are calculated based on the spatiotemporal evolution data of the temperature field and the refractive index distribution of the material. The energy deposition distribution, the spatiotemporal evolution data of the temperature field, and the migration trajectory and reconstructed morphology data of the phase transition interface are combined into a photothermal phase transition chain response structure.

6. The system according to claim 5, characterized in that, The chain-coupled response generation module is specifically used for: The material refractive index distribution is updated based on the reconstructed morphology data; The energy deposition distribution is recalculated based on the updated material refractive index distribution and the light spot polarization state; The spatiotemporal evolution data of the temperature field are recalculated based on the recalculated energy deposition distribution, the surface temperature sequence, and the material thermodynamic time-varying response parameters; The phase transition interface migration trajectory and morphology data are recalculated and reconstructed based on the recalculated spatiotemporal evolution data of the temperature field and the updated material refractive index distribution. Repeat the steps of updating the material refractive index distribution, recalculating the energy deposition distribution, recalculating the spatiotemporal evolution data of the temperature field, and recalculating the phase change interface migration trajectory and reconstructed morphology data until the root mean square error between two adjacent reconstructed morphology data is less than a preset threshold. Then, combine the final energy deposition distribution, spatiotemporal evolution data of the temperature field, and phase change interface migration trajectory and reconstructed morphology data into a photothermal phase change chain response structure.

7. The system according to claim 1, characterized in that, The deviation chain tracking module is specifically used for: Extract the noise frequency distribution and noise amplitude sequence from the residual phase noise of the laser source; The transfer functions of the light absorption stage, the thermal diffusion stage, and the phase change reconstruction stage are calculated based on the aforementioned photothermal phase change chain response structure. The noise frequency distribution is convolved with the transfer function of the light absorption stage to obtain a first stage deviation sequence; the first stage deviation sequence is convolved with the transfer function of the thermal diffusion stage to obtain a second stage deviation sequence; the second stage deviation sequence is convolved with the transfer function of the phase transition reconstruction stage to obtain a processing deviation chain.

8. The system according to claim 1, characterized in that, The trajectory adaptive control module is specifically used for: Obtain the spatial deviation component and phase deviation component in the machining deviation chain; Extract the major axis direction angle and ellipticity parameters of the polarization ellipse from the polarization state of the light spot; Obtain a preset reference scan trajectory; extract the tangent direction of each point from the reference scan trajectory; Calculate the angular deviation between the major axis direction angle and the tangent direction; The trajectory correction vector is calculated based on the included angle deviation value, the ellipticity parameter, the spatial deviation component, and the phase deviation component. The trajectory correction vector is superimposed on the position coordinates of each point on the reference scanning trajectory to obtain the corrected processing trajectory.

9. The system according to claim 1, characterized in that, The optical performance reverse optimization module is specifically used for: The refractive index gradient drift tensor and the response parameters of the photoinert region of the material are obtained from the twin state space; Extract the first and second components from the main diagonal elements of the refractive index gradient drift tensor, and use them as the first gradient component and the second gradient component, respectively. Obtain a preset target refractive index distribution curve, calculate the deviation between the first gradient component and the target refractive index distribution curve as the first refractive index deviation, and calculate the deviation between the second gradient component and the target refractive index distribution curve as the second refractive index deviation. The depth values ​​along the laser scanning direction, along the perpendicular scanning direction, and along the depth direction in the inert region boundary parameters are extracted from the photoinduced inert region response parameters of the material and used as the first depth value, the second depth value, and the third depth value, respectively. Extract the rate of change of refractive index of the inert region corresponding to the boundary parameters of the inert region, and use it as the deviation of the rate of change of refractive index; The adjustment amounts of the laser pulse energy and the scanning speed are calculated in reverse based on the first refractive index deviation, the second refractive index deviation, the refractive index change rate deviation, the first depth value, the second depth value, and the third depth value, and are output as the processing parameters after reverse adjustment.

10. The system according to claim 1, characterized in that, The twin state modeling module is also used for: Obtain the reverse-adjusted processing parameters output by the optical performance reverse optimization module; The reverse-adjusted processing parameters are input into the twin state space; The material's photo-induced inert region response parameters, refractive index gradient drift tensor, and thermodynamic time-varying response parameters are recalculated based on the reverse-adjusted processing parameters. The recalculated material refractive index distribution, the material photoinduced inert region response parameters, the refractive index gradient drift tensor, and the material thermodynamic time-varying response parameters are updated to the twin state space and output to the chain-coupled response generation module. The photothermal phase transition chain response structure is reconstructed based on the updated twin state space.

Images