Method for predicting and optimizing thermal coupling performance of off-axis beam expander optical machine

CN122652804APending Publication Date: 2026-08-28DALIAN UNIV OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611160889.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-03
Publication Date
2026-08-28

AI Technical Summary

Technical Problem

[0008]本发明旨在提供一种离轴扩束器光机热耦合性能预测与结构优化方法,以解决现有离轴扩束器优化设计中高保真光机热耦合仿真计算成本高、低保真模型精度不足、传统多保真代理模型难以处理非嵌套样本以及难以捕捉跨保真相关性局部变化的问题,实现离轴扩束器指向偏差和波前像差的快速、高精度预测与结构参数优化

Benefits of technology

(1)本发明通过建立高保真光机热耦合分析模型和低保真光机热耦合分析模型,并采用多输出多保真高斯过程进行多保真数据融合,能够在减少高保真仿真调用次数的同时提高目标高保真响应预测精度,降低离轴扩束器结构优化的计算成本。

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122652804A_ABST
    Figure CN122652804A_ABST
Patent Text Reader

Abstract

The present application belongs to the technical field of opto-mechanical thermal coupling analysis, optical system structure optimization, multi-fidelity proxy modeling and high-precision optoelectronic equipment design, and relates to an off-axis expander opto-mechanical thermal coupling performance prediction and structure optimization method. It comprises: establishing an off-axis expander parameterized structure model, constructing high and low fidelity opto-mechanical thermal coupling analysis models, obtaining the node displacement field of the primary mirror and the secondary mirror, decomposing the node displacement field into rigid body motion component and elastic deformation component, and calculating the pointing deviation and wavefront aberration to form a multi-fidelity training data set. Different fidelity responses are used as related tasks to construct a multi-output multi-fidelity Gaussian process proxy model. After training, the proxy model is used to predict the pointing deviation and wavefront aberration under the target high fidelity, and the off-axis expander structure parameters are optimized under the constraints of optical performance and structure size. The present application can reduce the number of high-fidelity simulations, improve the efficiency of opto-mechanical performance prediction and the accuracy of optimization design.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the technical fields of optomechanical-thermal coupling analysis, optical system structure optimization, multi-fidelity proxy modeling and high-precision optoelectronic equipment design, and specifically relates to a method for predicting the optomechanical-thermal coupling performance and optimizing the structure of an off-axis beam expander. Background Technology

[0002] Off-axis beam expanders are important optical components in remote sensing, space optoelectronic payloads, laser transmission, and high-precision optical measurement systems. Compared with coaxial beam expanders, off-axis beam expanders avoid central obstruction and have advantages such as foldable optical paths, compact structure, and high energy transmission efficiency, making them suitable for applications with high requirements for beam quality, wavefront quality, and pointing stability.

[0003] In actual service, off-axis beam expanders are typically subjected to a combination of factors, including gravity, temperature variations, support constraints, and assembly / adjustment errors. Due to the difference in thermal expansion coefficients between the mirror and the metal support structure, temperature changes cause thermal deformation of the support structure, changes in mirror orientation, and elastic deformation of the mirror, leading to increased beam pointing deviation and wavefront aberration. For high-precision off-axis beam expanders, pointing deviation and wavefront aberration are key indicators for evaluating their optomechanical-thermal coupling performance. Therefore, accurately predicting these optical performance responses and optimizing structural parameters during the structural design phase is of significant engineering importance.

[0004] Existing off-axis beam expander design methods typically rely on high-fidelity optomechanical-thermal coupled simulations. These methods require establishing a detailed finite element model, performing thermal-structural coupling analysis, extracting the displacement field of the mirror nodes, and then calculating pointing deviations and wavefront aberrations through ray tracing or optical analysis. While this method offers high accuracy, each calculation is time-consuming. During structural optimization, the optomechanical-thermal coupled simulation model needs to be repeatedly invoked, and the computational cost increases rapidly with the number of design variables and optimization iterations, making it difficult to meet the needs of rapid design optimization.

[0005] To reduce computational costs, coarse-mesh models, simplified physical models, or empirical response models are often used in engineering as low-fidelity analysis methods. Low-fidelity models are computationally fast and can provide a large number of auxiliary samples, but their prediction accuracy is usually lower than that of high-fidelity models, and they may not accurately reflect the local thermal deformation coupling relationship in certain structural parameter regions. If low-fidelity models are directly used for optical performance evaluation, the optimization results are likely to deviate from the true high-fidelity response.

[0006] Multi-fidelity surrogate modeling methods can integrate a small number of high-fidelity samples and a large number of low-fidelity samples, reducing the number of high-fidelity simulation calls while maintaining prediction accuracy. Existing multi-fidelity modeling methods typically assume a globally fixed correlation between responses of different fidelities, or require high-fidelity and low-fidelity samples to have nested, aligned, or fixed pairing relationships in the design variable space. However, the optomechanical-thermal coupled response of off-axis beam expanders exhibits significant locality and nonlinearity: in some design regions, the low-fidelity model can reflect the changing trend of the high-fidelity response well; in other design regions, due to the local compliance of the flexible hinge, changes in the stiffness of the secondary mirror support beam, or enhanced coupling from the thermal deformation of the mirror, the correlation between the low-fidelity and high-fidelity responses will be significantly reduced. Surrogate models that adopt the assumption of a globally fixed correlation are difficult to accurately describe this cross-fidelity correlation that varies with design variables.

[0007] Therefore, there is an urgent need for a method for predicting the optomechanical-thermal coupling performance and optimizing the structure of off-axis beam expanders. This method should be able to adaptively capture the correlation between high-fidelity and low-fidelity responses in different design regions under the conditions of a limited number of high-fidelity samples and non-nested and non-aligned low-fidelity and high-fidelity samples, so as to achieve high-precision prediction of pointing deviation and wavefront aberration and support the rapid optimization of off-axis beam expander structural parameters. Summary of the Invention

[0008] This invention aims to provide a method for predicting the optomechanical-thermal coupling performance and optimizing the structure of off-axis beam expanders, in order to solve the problems of high computational cost of high-fidelity optomechanical-thermal coupling simulation, insufficient accuracy of low-fidelity models, difficulty in handling non-nested samples by traditional multi-fidelity surrogate models, and difficulty in capturing local changes in cross-fidelity correlation in the optimization design of off-axis beam expanders. This method enables fast and high-precision prediction of pointing deviation and wavefront aberration of off-axis beam expanders and optimization of structural parameters.

[0009] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for predicting the optomechanical-thermal coupling performance and optimizing the structure of an off-axis beam expander, comprising the following steps: The first step is to establish a parameterized structural model of the off-axis beam expander.

[0010] The off-axis beam expander includes optical elements and a support structure for mounting the optical elements, specifically including a primary mirror, a secondary mirror, a secondary mirror support beam, a bidirectional flexible hinge, a support plate, and a support cylinder.

[0011] Based on the optical layout, support form, and thermally sensitive parts of the off-axis beam expander, structural dimensions that affect the thermal deformation coordination capability of the primary mirror, the support stability of the secondary mirror, and the optical performance are selected as design variables. This forms a design variable vector: (1) Among them, the design variable constraint domain for: (2) In the formula, For the first One structural design variable, and These represent the diameter and thickness of the flexible portion of the bidirectional flexible hinge, respectively. , and These are the cross-sectional height, inclination angle, and fixed end width of the secondary mirror support beam, respectively. and The first The lower and upper limits of each structural design variable.

[0012] The second step involves establishing high-fidelity optomechanical-thermal coupling analysis models and low-fidelity optomechanical-thermal coupling analysis models of the off-axis beam expander under the same design variables, temperature loads, gravity loads, and support constraints.

[0013] The high-fidelity optomechanical thermal coupling analysis model is used to obtain the optomechanical performance response with the target accuracy, while the low-fidelity optomechanical thermal coupling analysis model is used to obtain the auxiliary optomechanical performance response with lower computational cost. The optomechanical performance responses at different fidelities are expressed as follows: (3) In the formula, For fidelity level identification, Indicates low fidelity. Indicates high fidelity; Design variables; Indicates at the fidelity level The optomechanical performance response vector obtained under the corresponding optomechanical-thermal coupling analysis model; Indicates fidelity level The downward pointing deviation; Indicates fidelity level The wavefront aberration below.

[0014] The high-fidelity and low-fidelity optomechanical thermal coupling analysis models are determined based on mesh convergence error, response consistency index, and single-cycle computation time. Mesh convergence error is used to determine whether the model response after mesh refinement meets the preset accuracy requirements; the response consistency index is used to determine whether the response changes of the candidate low-fidelity optomechanical thermal coupling analysis model are consistent with those of the candidate high-fidelity optomechanical thermal coupling analysis model; and single-cycle computation time is used to determine whether the candidate model meets the preset computational cost threshold.

[0015] For different grid sizes Construct grid convergence error index for: (4) In the formula, This indicates the mesh size of the finite element model; and Indicates the dimensions of the two grids to be compared; Indicates the response type identifier; Indicates displacement response; Indicates stress response; Indicates the grid size is Time response type The corresponding response value; Indicates the grid size is Displacement response at time; Indicates the grid size is Stress response at time; To prevent positive numbers with a denominator of zero, when the convergence error corresponding to a mesh size is less than a preset convergence threshold, it is considered a candidate high-fidelity optomechanical-thermal coupling analysis model. The preset convergence threshold is used to determine whether the response change after mesh refinement meets the accuracy requirements of the high-fidelity model. For candidate low-fidelity optomechanical-thermal coupling analysis models, the response consistency index is used to characterize the degree of consistency between the candidate low-fidelity optomechanical-thermal coupling analysis model and the candidate high-fidelity optomechanical-thermal coupling analysis model in displacement and stress responses. Low-fidelity optomechanical-thermal coupling analysis models are selected from mesh sizes that meet the requirement of a single calculation time being lower than a preset computational cost threshold and a response consistency index reaching a preset retention threshold. The preset retention threshold refers to the minimum response consistency requirement that the candidate low-fidelity optomechanical-thermal coupling analysis model should meet relative to the candidate high-fidelity optomechanical-thermal coupling analysis model, and the preset computational cost threshold refers to the maximum allowable computation time for a single optomechanical-thermal coupling analysis.

[0016] The third step is to sample the design variable vector and then call the high-fidelity optomechanical-thermal coupling analysis model and the low-fidelity optomechanical-thermal coupling analysis model for analysis.

[0017] For each set of design variable samples, the same temperature load, gravity load, and support constraints are first applied to the corresponding fidelity model for thermo-structural coupling solution. Then, the nodal displacement fields of the primary and secondary mirror surfaces are extracted. The high-fidelity optomechanical-thermal coupling analysis model and the low-fidelity optomechanical-thermal coupling analysis model share the same design variable space and response definition. The difference between the two lies in the mesh size, the degree of local geometric simplification, the element type, or the solution accuracy settings. Based on the mirror surface geometry equations of the primary and secondary mirrors, the nodal displacement fields are decomposed into rigid body motion components and elastic deformation components. (5) In the formula, Indicates fidelity level The displacement field of the mirror nodal under the following conditions Indicates fidelity level The rigid body motion components below, Indicates fidelity level The elastic deformation component below.

[0018] Specifically, for the first For each mirror node, the node coordinates before and after deformation satisfy: (6) In the formula, and These are the fidelity levels. The node coordinates before and after deformation are shown below. For fidelity levels The rigid body rotation matrix is ​​as follows. For fidelity levels The rigid body translation vector below, For the first Each mirror node at the fidelity level The elastic deformation residuals are calculated. The rigid body rotation matrix and rigid body translation vector are determined using the least squares method. (7) In the formula, Let be the rigid body rotation matrix to be determined. Let be the rigid body translation vector to be determined.

[0019] The nodal pose change determined by the rigid body rotation matrix and rigid body translation vector is taken as the rigid body motion component, and the residual after subtracting the rigid body motion component from the total nodal displacement is taken as the elastic deformation component. The rigid body motion component is used to calculate the beam pointing deviation, and the elastic deformation component is used to calculate the wavefront aberration.

[0020] The fourth step is to calculate the pointing deviation of the off-axis beam expander by ray tracing based on the rigid body motion component; and to calculate the wavefront aberration of the off-axis beam expander by local coordinate transformation of the mirror and optical path difference based on the elastic deformation component. The pointing deviation and the wavefront aberration are then used as the optomechanical-thermal coupling performance response.

[0021] The calculation process for the pointing deviation includes: determining the starting point and direction of the incident ray; determining the normal vector of the deformed mirror surface based on the mirror's geometric equation and the corresponding rigid body motion components; sequentially solving for the intersection points of the ray with the primary and secondary mirror surfaces; calculating the direction of the ray reflected by the mirrors according to the law of reflection; and comparing the final outgoing ray direction with the ideal optical axis direction to obtain the pointing deviation of the off-axis beam expander. The mirrors include a primary mirror and a secondary mirror. When calculating the reflected ray direction of a single mirror surface according to the law of reflection, the reflection direction vector is: (8) In the formula, This represents the direction vector of the light ray after reflection by the current mirror surface. This represents the direction vector of the light ray incident on the current reflecting mirror surface. This represents the unit normal vector of the current reflecting mirror at the intersection of light rays. Based on the direction of the final outgoing light ray, the fidelity level... Pointing deviation below Represented as: (9) In the formula, The reference optical axis direction vector, For fidelity levels The final output ray direction vector is obtained by ray tracing.

[0022] The wavefront aberration calculation process includes: establishing local coordinate systems on the primary and secondary mirror surfaces respectively; transforming the elastic deformation components from the global coordinate system to the corresponding local coordinate system of the mirror surface; determining the optical path difference caused by mirror deformation based on the intersection points of the rays obtained from ray tracing and the mirror surface; and accumulating the optical path difference according to the aberration propagation relationship between the mirrors in the off-axis beam expander to obtain the wavefront aberration. The mapping between the local and global coordinate systems of the mirror surface is represented by a homogeneous coordinate transformation as follows: (10) In the formula, The coordinates are in the global coordinate system. These are the coordinates in the local coordinate system of the mirror surface. Let be the homogeneous transformation matrix from the global coordinate system to the local coordinate system of the mirror.

[0023] For discrete aperture sampling points, the root mean square value of wavefront aberration Represented as: (11) In the formula, Number of sampling points for aperture. For fidelity levels The next Optical path difference at each sampling point For fidelity levels The average optical path difference. In one embodiment, the optical path difference is expressed by Zernike polynomial expansion as: (12) In the formula, For normalized aperture radial coordinates, The aperture azimuth angle, For fidelity levels The optical path difference caused by the elastic deformation of the mirror surface. The number of Zernike polynomial terms involved in the expansion. For the first Zernike polynomials, For fidelity levels Next The expansion coefficients of the Zernike polynomial.

[0024] The fifth step involves constructing a multi-fidelity training dataset, including high-fidelity samples and low-fidelity samples, using the design variable vector as input and the pointing deviation and wavefront aberration as output.

[0025] The number of high-fidelity samples and low-fidelity samples are allowed to differ, and their sampling positions in the design variable space are allowed to be non-overlapping. The multi-fidelity training dataset is represented as follows: (13) (14) In the formula, For multi-fidelity training datasets, For fidelity levels The following training data subset, For fidelity levels Next Design variable vectors for each sample For fidelity levels Next The optomechanical-thermal coupling performance response vector of each sample To ensure the number of tasks, For fidelity levels The number of samples at different fidelity levels. Nested sampling or aligned sampling is not required between tasks of different fidelity levels, and the number of low-fidelity samples is greater than or equal to the number of high-fidelity samples.

[0026] Before constructing the multi-fidelity training dataset, the design variables and the optomechanical performance response are normalized: (15) In the formula, For the normalized first One design variable, The first before normalization One design variable, For the first The lower bound of a design variable. For the first The upper limit of a design variable; Fidelity levels before normalization Next Each optomechanical thermal coupling performance response component For normalized fidelity levels Next Each optomechanical thermal coupling performance response component; among which... For the response component numbering, Indicates pointing deviation. Indicates wavefront aberration; and These are the fidelity levels. Next The mean and standard deviation of the response components of the optomechanical thermal coupling performance.

[0027] Step 6: Construct a multi-output, multi-fidelity Gaussian process prediction model.

[0028] Using the design variable vector as input and the optomechanical-thermal coupling performance response vector, including pointing deviation and wavefront aberration, as output, the responses at different fidelities are used as relevant tasks in the multi-output multi-fidelity Gaussian process prediction model. Multiple fidelity tasks share the same design variable space; information is transferred between different fidelity tasks through a cross-fidelity covariance function, allowing low-fidelity samples to supplement the information loss caused by insufficient high-fidelity samples, while preserving the corrective effect of high-fidelity samples on the target optomechanical performance response.

[0029] In the multi-output multi-fidelity Gaussian process prediction model, the correlation between tasks of different fidelity is represented as a weighted superposition of several potential correlation components; each potential correlation component is described by a potential Gaussian process, and the input correlation gating function is used to determine the contribution of each potential correlation component at different design variable locations. Fidelity hierarchy The corresponding latent response function is expressed as: (16) In the formula, For fidelity levels The corresponding potential response function; The number of potentially relevant components; Number the potentially relevant components; For the first The latent Gaussian process corresponding to each latent correlation component is used to describe the response as a function of design variables under that latent correlation component. The changing relationship; For fidelity levels In the The task weights under each potentially relevant component are used to characterize the response strength of the fidelity level to that potentially relevant component. For the first One potentially relevant component in the design variables The gating weight at a given position is used to characterize the contribution of the potential correlation component to the cross-fidelity correlation at the current position.

[0030] The seventh step is to introduce an input correlation gating function into the multi-output multi-fidelity Gaussian process proxy model. The input correlation gating function is used to adjust the contribution of different potential correlation components to cross-fidelity information transmission, so that the correlation between the optomechanical-thermal coupling performance responses of different fidelities can be adaptively adjusted as the position of the design variable vector changes.

[0031] In a preferred embodiment, the input correlation gating function includes gating weights for two potentially correlated components, expressed as: (17) in, The function is: (18) In the formula, This is the gating coefficient vector. For bias parameters, For design variable vector Gating coefficient vector and bias parameters Determined gating function input variables, and These represent the first and second latent correlation components in the design variable vector, respectively. The weights at each point. The gating coefficient vector. The bias parameter is used to characterize the orientation of design variables that affect the spatial variation of cross-fidelity correlation. Used to control the transition positions between different potentially related components.

[0032] Step 8: Construct a variable-related cross-fidelity covariance function based on the input-related gating function, task mode weights, and input space kernel function, and form a unified covariance matrix by covariance blocks between tasks of different fidelity, and perform fusion modeling on the multi-fidelity training dataset.

[0033] The variable-related cross-fidelity covariance function Represented as: (19) In the formula, and For fidelity level identification, and ,in, Indicates a low-fidelity task. Indicates a high-fidelity task; when When, it represents the covariance between samples within the same fidelity level. When, it represents the cross-fidelity covariance between low-fidelity tasks and high-fidelity tasks; and They represent the first The and the first A sample of design variables, Indicates the first Gating weights of each potentially relevant component, and Indicates different fidelity tasks in the first... Task pattern weights of each potentially relevant component Indicates the first Gaussian process kernel function for each potentially correlated component.

[0034] In a preferred embodiment, the latent Gaussian process kernel function Using the quadratic exponential kernel function: (20) In the formula, For the first Amplitude parameters of a potential Gaussian process, For the first The potential Gaussian process in the th... Length-scale related parameters in the direction of each design variable; Indicates the first The design variable sample in the _th _ ... The components in the direction of each design variable, Indicates the first The design variable sample in the _th _ ... The component in the direction of each design variable.

[0035] The samples from all fidelity tasks are arranged in blocks according to fidelity to obtain a unified covariance matrix. : (twenty one) In the formula, For high-fidelity samples, covariance blocks For the covariance block between low-fidelity samples, and For cross-fidelity covariance blocks between high-fidelity and low-fidelity samples, It is the identity matrix. To observe the noise variance.

[0036] Therefore, the cross-fidelity covariance is no longer a global constant within the design space, but rather a variable correlation function adjusted by the input correlation gating function. In design regions where the low-fidelity and high-fidelity responses are highly consistent, the model increases the contribution of low-fidelity samples to high-fidelity predictions; in design regions where the low-fidelity and high-fidelity responses differ significantly, the model reduces the misleading effect of low-fidelity samples, thereby improving the accuracy of the target high-fidelity response prediction.

[0037] The ninth step is to determine the hyperparameters of the multi-output multi-fidelity Gaussian process prediction model by maximizing the marginal likelihood function, thereby obtaining the trained multi-output multi-fidelity Gaussian process prediction model and its prediction components for pointing bias and wavefront aberration.

[0038] The hyperparameters include input space kernel function parameters, input correlation gating function parameters, task mode weights, and observation noise parameters. These hyperparameters are obtained by maximizing the marginal log-likelihood function. (twenty two) In the formula, This includes potential kernel function parameters, gating coefficient vector, bias parameters, task weights, and noise parameters. The hyperparameter estimates that maximize the marginal log-likelihood function; The response vector of the training sample; The total number of training samples is denoted as . The hyperparameters are globally searched using a genetic algorithm and locally optimized using a sequential least squares programming algorithm. The genetic algorithm is used to perform global exploration in the hyperparameter space, reducing the probability of getting trapped in local optima; the sequential least squares programming algorithm is used to locally refine the candidate solutions obtained by the genetic algorithm, improving the convergence accuracy of the marginal likelihood function optimization.

[0039] Step 10: Input the design variables to be optimized into the trained multi-output multi-fidelity Gaussian process prediction model to predict the pointing deviation and wavefront aberration under the target high-fidelity task. Then, optimize the design variables with the goal of reducing pointing deviation and with the wavefront aberration threshold and the range of design variable values ​​as constraints, to obtain the optimized off-axis beam expander structural parameters.

[0040] For the design variables to be predicted The predicted mean of the target high-fidelity response and prediction variance They are respectively: (twenty three) (twenty four) In the formula, Let be the covariance vector between the sample to be predicted and all training samples. The optimized off-axis beam expander structural parameters are expressed as follows: (25) In the formula, and These represent the pointing bias and wavefront aberration predicted under a high-fidelity target mission, respectively. To preset the wavefront aberration constraint threshold, For the first The lower bound of a design variable. For the first The upper limit of a design variable.

[0041] After obtaining the optimized off-axis beam expander structural parameters, these parameters are substituted back into the high-fidelity optomechanical-thermal coupling analysis model. The pointing deviation and wavefront aberration are then recalculated following the same process: nodal displacement field extraction, separation of rigid body motion and elastic deformation components, ray tracing, and wavefront aberration calculation. When the verification results meet the preset optical performance constraints, the optimized structural parameters are output. If the verification results do not meet the preset optical performance constraints, the high-fidelity verification sample is added to the multi-fidelity training dataset, and the multi-output multi-fidelity Gaussian process prediction model is retrained for further optimization.

[0042] The beneficial effects of this invention are as follows: (1) By establishing a high-fidelity optomechanical-thermal coupling analysis model and a low-fidelity optomechanical-thermal coupling analysis model, and using a multi-output multi-fidelity Gaussian process for multi-fidelity data fusion, this invention can improve the accuracy of target high-fidelity response prediction while reducing the number of high-fidelity simulation calls, and reduce the computational cost of off-axis beam expander structure optimization.

[0043] (2) This invention does not require high-fidelity samples and low-fidelity samples to be nested or aligned, and can be applied to actual engineering data with a small number of high-fidelity samples, a large number of low-fidelity samples and inconsistent sampling positions, thereby improving the utilization rate of multi-source simulation samples.

[0044] (3) This invention constructs the cross-fidelity covariance of variables by inputting the relevant gating function, so that the correlation between responses of different fidelity can adapt to the position of the design variables. This overcomes the shortcomings of the global fixed task correlation of the traditional multi-output Gaussian process model and can more accurately describe the local thermal deformation coupling relationship in different design regions of the off-axis beam expander.

[0045] (4) The present invention decomposes the displacement field of the mirror node into rigid motion components and elastic deformation components, which are used for pointing deviation and wavefront aberration calculation, respectively, so that the finite element thermal structure response can establish a clear correspondence with the optical performance index, and improves the physical consistency of the input and output data of the proxy model.

[0046] (5) The present invention can simultaneously predict pointing deviation and wavefront aberration, and use the two to optimize the structural parameters of off-axis beam expanders, thereby achieving a more reasonable design balance between the thermal deformation coordination capability of the primary mirror, the support stiffness of the secondary mirror and the overall optical performance. Attached Figure Description

[0047] Figure 1 This is an overall flowchart of the method of the present invention; Figure 2 This is a schematic diagram of the overall structure of an off-axis beam expander. In the diagram, 1 is the primary mirror, 2 is the secondary mirror, 3 is the secondary mirror support beam, 4 is the bidirectional flexible hinge, 5 is the support plate, and 6 is the support cylinder. Figure 3 Schematic diagram for constructing high-fidelity and low-fidelity optomechanical-thermal coupling analysis models; Figure 4 Flowchart for decomposition of nodal displacement field and extraction of optomechanical properties of mirror; Figure 5 Schematic diagram of the structure of a multi-output, multi-fidelity Gaussian process prediction model; Figure 6 This is a flowchart of the off-axis beam expander structure optimization based on the surrogate model. Detailed Implementation

[0048] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings and technical solutions.

[0049] The overall process of the off-axis beam expander optomechanical-thermal coupling performance prediction and structural optimization method of the present invention is as follows: Figure 1 As shown. Figure 2 As shown, the off-axis beam expander includes a primary mirror 1, a secondary mirror 2, a secondary mirror support beam 3, a bidirectional flexible hinge 4, a support plate 5, and a support cylinder 6. The primary mirror 1 and secondary mirror 2 are off-axis parabolic mirrors, and the optical path adopts an off-axis reflection layout without central obstruction. The primary mirror 1 is mounted on the support plate 5, and the support plate 5 is mounted on the support cylinder 6. The primary mirror 1 and the secondary mirror support beam 3 are connected by a three-point bidirectional flexible hinge 4. The bidirectional flexible hinge 4 has a certain degree of compliance in both the radial and axial directions, used to alleviate the thermal deformation mismatch caused by the difference in the coefficients of thermal expansion between the primary mirror 1 and the support plate 5. The secondary mirror 2 is located at the end of the inclined secondary mirror support beam 3, which adopts a variable cross-section or rectangular tube structure to improve bending stiffness.

[0050] The off-axis beam expander structure optimization process based on the surrogate model is as follows: Figure 6 As shown.

[0051] In this embodiment, the design variable vector is: (26) in, The diameter of the bidirectional flexible hinge is... The thickness of the flexible part of the two-way flexible hinge. The height of the secondary mirror support beam section. The tilt angle of the secondary mirror support beam. This refers to the width of the fixed end of the secondary mirror support beam. The range of values ​​for the design variables can be determined based on structural space, manufacturing constraints, and optical performance requirements. Table 1 shows the design variables and their value ranges in this embodiment.

[0052] Table 1 Design variables and their value ranges

[0053] A parametric finite element model of the off-axis beam expander was established, modeling the primary mirror, secondary mirror, support beam, flexible hinge, support plate, and support cylinder. Fixed support constraints were applied to the mirror mount flange surface, along with gravity and temperature loads. The temperature load can be set according to actual operating conditions; in one embodiment, the ambient temperature was increased from 20°C to 25°C. For each set of design variable samples, a thermal-structural coupling analysis was first performed to obtain the nodal displacement fields of the primary and secondary mirrors. The decomposition process of the mirror nodal displacement field and the extraction of optomechanical performance are as follows: Figure 4 As shown.

[0054] To construct a multi-fidelity data source, a mesh convergence analysis was performed on the finite element model of the off-axis beam expander. Maximum displacement and maximum equivalent stress were used as convergence evaluation indicators to compare response changes and computation time at different mesh scales. When the response change was less than a preset threshold after further mesh refinement, this mesh model was selected as the high-fidelity optomechanical-thermal coupling analysis model. Simultaneously, a low-fidelity optomechanical-thermal coupling analysis model was selected from coarse mesh models that met the computational cost threshold and whose response consistency index reached a preset retention threshold. The high-fidelity and low-fidelity optomechanical-thermal coupling analysis models were constructed as follows: Figure 3 As shown. In one embodiment, the high-fidelity optomechanical-thermal coupling analysis model uses a finite element mesh with a maximum element size of 4 mm, while the low-fidelity optomechanical-thermal coupling analysis model uses a finite element mesh with a maximum element size of 20 mm. This example is only a preferred implementation; the high-fidelity and low-fidelity models can also be obtained by setting different element types, different degrees of local geometric simplification, or different solution precisions.

[0055] The nodal displacement fields of the primary and secondary mirrors are separated into rigid body motion components and elastic deformation components. Let the initial set of nodal coordinates of the reflecting mirrors be... The set of node coordinates after finite element analysis is Then, the node deformation can be decomposed into global translation, global rotation, and local elastic deformation. Based on the mirror surface elevation function... Establish a nodal fitting error function, and obtain the rigid body translation and rotation parameters using the least squares method. The fitting objective is expressed as: (27) The nodal pose changes determined by the rigid body rotation matrix and rigid body translation vector are taken as rigid body motion components, and the residual after subtracting the rigid body motion components from the total nodal displacement is taken as elastic deformation components.

[0056] In the calculation of pointing deviation, the origin and direction of the incident ray are defined, and the intersection point of the ray and the mirror is calculated based on the normal vector of the deformed reflecting mirror. For the reflecting surface, the direction of the reflected ray is calculated according to the law of reflection: (28) In the formula, Let be the incident vector. The unit normal vector of the mirror surface. The reflection vector is used. The direction of the final emitted ray is compared with the ideal optical axis direction to obtain the pointing deviation of the off-axis beam expander.

[0057] In wavefront aberration calculations, local coordinate systems are established on the primary and secondary mirror surfaces. The origin of the local coordinate system is taken as the vertex of the mirror surface. The axis is along the direction of the optical principal axis. The axis is determined based on the directions of the incident and reflected rays. The axis is determined by the right-hand rule. The elastic deformation of the mirror in the global coordinate system is mapped to the local coordinate system through homogeneous coordinate transformation, and the optical path difference caused by the mirror deformation is calculated based on the intersection points of the light rays and the mirror. The optical path differences caused by different mirrors are accumulated according to the light path propagation relationship to obtain the system wavefront aberration. Wavefront aberration can be characterized by the root mean square (RMS) value, peak-to-valley (PV) value, or Zernike polynomial coefficients.

[0058] After obtaining high-fidelity pointing bias, high-fidelity wavefront aberration, low-fidelity pointing bias, and low-fidelity wavefront aberration under different design variable samples, a multi-fidelity training dataset is constructed. A smaller number of high-fidelity samples and a larger number of low-fidelity samples can be used, and a one-to-one correspondence between high-fidelity and low-fidelity samples in the design variable space is not required. Sampling methods can include Latin hypercube sampling, orthogonal sampling, uniform sampling, or random sampling.

[0059] A multi-output, multi-fidelity Gaussian process prediction model was established. The design variable vector was used as input, and the optomechanical-thermal coupling performance response vector, including pointing bias and wavefront aberration, was used as output. High-fidelity pointing bias response, low-fidelity pointing bias response, high-fidelity wavefront aberration response, and low-fidelity wavefront aberration response were all considered as related tasks in the model, and information fusion between different tasks was achieved through a variable-correlation cross-fidelity covariance function.

[0060] For any design variable sample First, calculate the input-related gating function. and The gating function determines the weights of different latent correlation components based on the location of the design variable. When the design variable is located in a region where the low-fidelity model and the high-fidelity model are highly correlated, the gating function increases the contribution of the corresponding latent correlation component; when the design variable is located in a region where the low-fidelity model has a large error, the gating function reduces the impact of the low-fidelity response on the target high-fidelity prediction.

[0061] Then, a unified covariance matrix is ​​constructed based on the gating function, task mode weights, and input space kernel function. Each covariance block in the unified covariance matrix corresponds to a correlation between a pair of fidelity tasks. Since the covariance blocks are regulated by the input correlation gating function, the unified covariance matrix can characterize the local variation of cross-fidelity correlation within the design variable space.

[0062] The structure of a multi-output multi-fidelity Gaussian process prediction model is as follows: Figure 5 As shown.

[0063] During the training phase, the hyperparameters of the multi-output multi-fidelity Gaussian process prediction model are estimated by maximizing the marginal likelihood function. To improve training stability, a genetic algorithm is first used to perform a global search on the input space kernel function parameters, gating function parameters, task mode weights, and observation noise parameters; then, a sequential least squares programming algorithm is used to locally optimize the candidate solutions, resulting in the trained multi-output multi-fidelity Gaussian process prediction model.

[0064] After training, the design variables to be evaluated are input into the multi-output multi-fidelity Gaussian process prediction model to obtain the predicted values ​​of pointing deviation and wavefront aberration under the target high-fidelity task. Based on these prediction results, an off-axis beam expander structure optimization model is established. The optimization problem can be expressed as: (29) In the formula, The predicted pointing deviation value for a high-fidelity target task. The predicted root mean square value of wavefront aberration for a high-fidelity target task. To preset the wavefront aberration threshold, and The first The lower and upper bounds of each design variable are defined. The optimization algorithm can employ genetic algorithms, particle swarm optimization, sequential quadratic programming, or other constrained optimization algorithms.

[0065] In one embodiment, the design variables are optimized with the goal of reducing pointing deviation and the constraint that the root mean square value of wavefront aberration does not exceed 80 nm. A genetic algorithm is used to solve the optimization problem, obtaining the optimized design variables. The optimization results are then substituted back into a high-fidelity optomechanical-thermal coupling analysis model for verification. Table 2 shows a comparison of the structural parameters and optical performance before and after optimization in one embodiment.

[0066] Table 2 Comparison of structural parameters and optical performance before and after optimization.

[0067] As shown in Table 2, in this embodiment, the pointing deviation of the optimized off-axis beam expander is reduced from 6.900 μrad to 3.891 μrad, and the root mean square value of the wavefront aberration is reduced from 89.620 nm to 77.681 nm. The optimized structure can achieve a more reasonable balance between flexibility and support stiffness under thermal load, thereby reducing the impact of primary mirror thermal deformation mismatch and secondary mirror support deformation on optical performance.

[0068] The method described in this invention is not limited to the number of high-fidelity and low-fidelity models mentioned above. When there are more than two sources of fidelity, each source of fidelity can be treated as a task in a multi-output multi-fidelity Gaussian process proxy model, and a unified covariance matrix can be constructed in the same manner. This invention is also not limited to predicting only pointing deviation and wavefront aberration; in off-axis beam expander design, mirror pose error, mirror surface shape error, structural mass, maximum stress, or other responses related to optomechanical-thermal coupling performance can also be used as additional outputs for extended modeling.

Claims

1. A method for predicting the optomechanical-thermal coupling performance and optimizing the structure of an off-axis beam expander, characterized in that, The steps are as follows: The first step is to establish a parameterized structural model of the off-axis beam expander; Structural dimensions that affect the thermal deformation compatibility of the primary mirror, the support stability of the secondary mirror, and optical performance are selected as design variables. This forms a design variable vector: (1) Among them, the design variable constraint domain for: (2) In the formula, For the first One structural design variable, and These represent the diameter and thickness of the flexible portion of the bidirectional flexible hinge, respectively. , and These are the cross-sectional height, inclination angle, and fixed end width of the secondary mirror support beam, respectively. and The first The lower and upper limits of each structural design variable; The second step is to establish a high-fidelity optomechanical-thermal coupling analysis model and a low-fidelity optomechanical-thermal coupling analysis model of the off-axis beam expander under the same design variables, the same temperature load, the same gravity load and the same support constraints. The third step is to sample the design variable vector and then call the high-fidelity optomechanical-thermal coupling analysis model and the low-fidelity optomechanical-thermal coupling analysis model for analysis respectively. The fourth step is to calculate the pointing deviation of the off-axis beam expander by ray tracing based on the rigid body motion component; and to calculate the wavefront aberration of the off-axis beam expander by local coordinate transformation of the mirror and optical path difference based on the elastic deformation component. The pointing deviation and the wavefront aberration are then used as the optomechanical-thermal coupling performance response. The fifth step involves constructing a multi-fidelity training dataset, including high-fidelity samples and low-fidelity samples, using the design variable vector as input and the pointing deviation and wavefront aberration as output. Step 6: Construct a multi-output, multi-fidelity Gaussian process prediction model; The design variable vector is used as input, and the optomechanical thermal coupling performance response vector, including pointing deviation and wavefront aberration, is used as output. The responses at different fidelities are used as related tasks in the multi-output multi-fidelity Gaussian process prediction model. Multiple fidelity tasks share the same design variable space. Information is transferred between different fidelity tasks through the cross-fidelity covariance function, so that low-fidelity samples can supplement the information loss caused by insufficient number of high-fidelity samples, while retaining the corrective effect of high-fidelity samples on the target optomechanical performance response. The seventh step is to introduce an input correlation gate function into the multi-output multi-fidelity Gaussian process proxy model. The input correlation gate function is used to adjust the contribution of different potential correlation components to cross-fidelity information transmission, so that the correlation between the optomechanical-thermal coupling performance responses of different fidelities can be adaptively adjusted as the position of the design variable vector changes. Step 8: Construct a variable-related cross-fidelity covariance function based on the input-related gating function, task mode weights, and input space kernel function, and form a unified covariance matrix by covariance blocks between tasks of different fidelity, and perform fusion modeling on the multi-fidelity training dataset; The ninth step is to determine the hyperparameters of the multi-output multi-fidelity Gaussian process prediction model by maximizing the marginal likelihood function, thereby obtaining the trained multi-output multi-fidelity Gaussian process prediction model and its prediction components for pointing bias and wavefront aberration. Step 10: Input the design variables to be optimized into the trained multi-output multi-fidelity Gaussian process prediction model to predict the pointing deviation and wavefront aberration under the target high-fidelity task. Then, optimize the design variables with the goal of reducing pointing deviation and with the wavefront aberration threshold and the range of design variable values ​​as constraints, to obtain the optimized off-axis beam expander structural parameters. After obtaining the optimized off-axis beam expander structural parameters, the optimized structural parameters are substituted back into the high-fidelity optomechanical-thermal coupling analysis model, and the pointing deviation and wavefront aberration are recalculated according to the same nodal displacement field extraction, rigid body motion component and elastic deformation component separation, ray tracing and wavefront aberration calculation process. When the verification result meets the preset optical performance constraints, the optimized structural parameters are output. When the verification result does not meet the preset optical performance constraints, the high-fidelity verification sample is added to the multi-fidelity training dataset, and the multi-output multi-fidelity Gaussian process prediction model is retrained for further optimization.

2. The method for predicting the optomechanical-thermal coupling performance and optimizing the structure of an off-axis beam expander according to claim 1, characterized in that, In the second step, The high-fidelity optomechanical thermal coupling analysis model is used to obtain the optomechanical performance response with the target accuracy, and the low-fidelity optomechanical thermal coupling analysis model is used to obtain the auxiliary optomechanical performance response with lower computational cost; the optomechanical performance response at different fidelities is expressed as follows: (3) In the formula, For fidelity level identification, Indicates low fidelity. Indicates high fidelity; For design variables; Indicates at the fidelity level The optomechanical performance response vector obtained under the corresponding optomechanical-thermal coupling analysis model; Indicates fidelity level The downward pointing deviation; Indicates fidelity level Wavefront aberrations; The high-fidelity optomechanical thermal coupling analysis model and the low-fidelity optomechanical thermal coupling analysis model are determined based on mesh convergence error, response consistency index, and single computation time. Mesh convergence error is used to determine whether the model response after mesh refinement meets the preset accuracy requirements; the response consistency index is used to determine whether the response change of the candidate low-fidelity optomechanical thermal coupling analysis model is consistent with that of the candidate high-fidelity optomechanical thermal coupling analysis model; and single computation time is used to determine whether the candidate model meets the preset computation cost threshold. For different grid sizes Construct grid convergence error index for: (4) In the formula, This indicates the mesh size of the finite element model; and Indicates the dimensions of the two grids to be compared; Indicates the response type identifier; Indicates displacement response; Indicates stress response; Indicates the grid size is Time response type The corresponding response value; Indicates the grid size is Displacement response at time; Indicates the grid size is Stress response at time; To prevent positive numbers with zero denominators; when the convergence error corresponding to a mesh size is less than a preset convergence threshold, it is used as a candidate high-fidelity optomechanical-thermal coupling analysis model; for candidate low-fidelity optomechanical-thermal coupling analysis models, the response consistency index is used to characterize the degree of consistency between the candidate low-fidelity optomechanical-thermal coupling analysis model and the candidate high-fidelity optomechanical-thermal coupling analysis model in displacement response and stress response; among the mesh sizes that meet the requirements of single calculation time being lower than a preset calculation cost threshold and response consistency index reaching a preset retention threshold, the low-fidelity optomechanical-thermal coupling analysis model is selected; the preset calculation cost threshold refers to the maximum allowable calculation time for a single optomechanical-thermal coupling analysis.

3. The method for predicting the optomechanical-thermal coupling performance and optimizing the structure of an off-axis beam expander according to claim 1, characterized in that, The third step is as follows: For each set of design variable samples, the same temperature load, gravity load, and support constraints are first applied to the corresponding fidelity model for thermo-structural coupling solution. Then, the nodal displacement fields of the primary and secondary mirror surfaces are extracted. The high-fidelity optomechanical-thermal coupling analysis model and the low-fidelity optomechanical-thermal coupling analysis model share the same design variable space and response definition. Based on the mirror surface geometry equations of the primary and secondary mirrors, the nodal displacement fields are decomposed into rigid body motion components and elastic deformation components. (5) In the formula, Indicates fidelity level The displacement field of the mirror nodal under the following conditions Indicates fidelity level The rigid body motion components below, Indicates fidelity level The elastic deformation component below; Specifically, for the first For each mirror node, the node coordinates before and after deformation satisfy: (6) In the formula, and These are the fidelity levels. The node coordinates before and after deformation are shown below. For fidelity levels The rigid body rotation matrix is ​​as follows. For fidelity levels The rigid body translation vector below, For the first Each mirror node at the fidelity level The elastic deformation residuals are as follows; the rigid body rotation matrix and rigid body translation vector are determined by least squares: (7) In the formula, Let be the rigid body rotation matrix to be determined. Let be the rigid body translation vector to be determined; The nodal pose change determined by the rigid body rotation matrix and the rigid body translation vector is taken as the rigid body motion component, and the residual after deducting the rigid body motion component from the total nodal displacement is taken as the elastic deformation component.

4. The method for predicting the optomechanical-thermal coupling performance and optimizing the structure of an off-axis beam expander according to claim 1, characterized in that, In the fourth step, The calculation process for the pointing deviation includes: determining the starting point and direction of the incident ray; determining the normal vector of the deformed mirror surface based on the mirror's geometric equation and the corresponding rigid body motion components; sequentially solving for the intersection points of the ray with the primary and secondary mirror surfaces; calculating the direction of the ray reflected by the mirrors according to the law of reflection; comparing the final outgoing ray direction with the ideal optical axis direction to obtain the pointing deviation of the off-axis beam expander; wherein, the mirror includes a primary mirror and a secondary mirror; when calculating the reflected ray direction of a single mirror surface according to the law of reflection, the reflection direction vector is: (8) In the formula, This represents the direction vector of the light ray after reflection by the current mirror surface. This represents the direction vector of the light ray incident on the current reflecting mirror surface. This represents the unit normal vector of the current reflecting mirror at the intersection of light rays; the fidelity level is based on the direction of the final outgoing light ray. The pointing deviation below Represented as: (9) In the formula, The reference optical axis direction vector, For fidelity levels The final output ray direction vector obtained by ray tracing; The wavefront aberration calculation process includes: establishing local coordinate systems on the primary and secondary mirror surfaces respectively; transforming the elastic deformation components from the global coordinate system to the corresponding local coordinate system of the mirror surface; determining the optical path difference caused by mirror deformation based on the intersection points of the rays obtained from ray tracing and the mirror surface; accumulating the optical path difference according to the aberration propagation relationship between the mirrors in the off-axis beam expander to obtain the wavefront aberration; the mapping between the local and global coordinate systems of the mirror surface is expressed using a homogeneous coordinate transformation as follows: (10) In the formula, The coordinates are in the global coordinate system. These are the coordinates in the local coordinate system of the mirror surface. This is the homogeneous transformation matrix from the global coordinate system to the local coordinate system of the mirror. For discrete aperture sampling points, the root mean square value of wavefront aberration Represented as: (11) In the formula, Number of sampling points for aperture. For fidelity levels The next Optical path difference at each sampling point For fidelity levels The average optical path difference.

5. The method for predicting the optomechanical-thermal coupling performance and optimizing the structure of an off-axis beam expander according to claim 1, characterized in that, In the fifth step, the multi-fidelity training dataset is represented as follows: (13) (14) In the formula, For multi-fidelity training datasets, For fidelity levels The following training data subset, For fidelity levels Next Design variable vectors for each sample For fidelity levels Next The optomechanical-thermal coupling performance response vector of each sample To ensure the number of tasks, For fidelity levels The number of samples below; Before constructing the multi-fidelity training dataset, the design variables and the optomechanical performance response are normalized: (15) In the formula, For the normalized first One design variable, The first before normalization One design variable, For the first The lower bound of a design variable. For the first The upper limit of a design variable; Fidelity levels before normalization Next Each optomechanical thermal coupling performance response component For normalized fidelity levels Next Each optomechanical thermal coupling performance response component; among which... For the response component numbering, Indicates pointing deviation. Indicates wavefront aberration; and These are the fidelity levels. Next The mean and standard deviation of the response components of the optomechanical thermal coupling performance.

6. The method for predicting the optomechanical-thermal coupling performance and optimizing the structure of an off-axis beam expander according to claim 1, characterized in that, In the sixth step, In the multi-output multi-fidelity Gaussian process prediction model, the correlation between tasks of different fidelity is represented as a weighted superposition of several potential correlation components; each potential correlation component is described by a potential Gaussian process, and the input correlation gating function is used to determine the contribution of each potential correlation component at different design variable locations; fidelity hierarchy The corresponding latent response function is expressed as: (16) In the formula, For fidelity levels The corresponding potential response function; The number of potentially relevant components; Number the potentially relevant components; For the first The latent Gaussian process corresponding to each latent correlation component is used to describe the response as a function of design variables under that latent correlation component. The changing relationship; For fidelity levels In the The task weights under each potentially relevant component are used to characterize the response strength of the fidelity level to that potentially relevant component. For the first One potentially relevant component in the design variables The gating weight at a given position is used to characterize the contribution of the potential correlation component to the cross-fidelity correlation at the current position.

7. The method for predicting the optomechanical-thermal coupling performance and optimizing the structure of an off-axis beam expander according to claim 1, characterized in that, In step seven, the input correlation gating function includes gating weights for two potential correlation components, expressed as: (17) in, The function is: (18) In the formula, This is the gating coefficient vector. For bias parameters, For design variable vector Gating coefficient vector and bias parameters Given a defined gating function input variable, and These represent the first and second latent correlation components in the design variable vector, respectively. Weight at each location.

8. The method for predicting the optomechanical-thermal coupling performance and optimizing the structure of an off-axis beam expander according to claim 1, characterized in that, In the eighth step, The variable-related cross-fidelity covariance function Represented as: (19) In the formula, and For fidelity level identification, and ,in, Indicates a low-fidelity task. Indicates a high-fidelity task; when When, it represents the covariance between samples within the same fidelity level. When, it represents the cross-fidelity covariance between low-fidelity tasks and high-fidelity tasks; and They represent the first The and the first A sample of design variables, Indicates the first Gating weights of potential relevant components, and Indicates different fidelity tasks in the first... Task pattern weights of each potentially relevant component Indicates the first Gaussian process kernel function with potential correlated components; The samples from all fidelity tasks are arranged in blocks according to fidelity to obtain a unified covariance matrix. : (21) In the formula, For high-fidelity samples, covariance blocks For the covariance block between low-fidelity samples, and For cross-fidelity covariance blocks between high-fidelity and low-fidelity samples, It is the identity matrix. To observe the noise variance.

9. The method for predicting the optomechanical-thermal coupling performance and optimizing the structure of an off-axis beam expander according to claim 1, characterized in that, In the ninth step, The hyperparameters include input space kernel function parameters, input correlation gating function parameters, task mode weights, and observation noise parameters; the hyperparameters are obtained by maximizing the marginal log-likelihood function. (22) In the formula, This includes potential kernel function parameters, gating coefficient vector, bias parameters, task weights, and noise parameters. The hyperparameter estimates that maximize the marginal log-likelihood function; The response vector of the training sample; This represents the total number of training samples.

10. The method for predicting the optomechanical-thermal coupling performance and optimizing the structure of an off-axis beam expander according to claim 1, characterized in that, In the tenth step, For the design variables to be predicted The predicted mean of the target high-fidelity response and prediction variance They are respectively: (23) (24) In the formula, Let be the covariance vector between the sample to be predicted and all training samples; the optimized off-axis beam expander structural parameters are expressed as: (25) In the formula, and These represent the pointing bias and wavefront aberration predicted under a high-fidelity target mission, respectively. To preset the wavefront aberration constraint threshold, For the first The lower bound of a design variable. For the first The upper limit of a design variable.