Liquid lens performance optimization method and system

Abstract

The present invention discloses a liquid lens performance optimization method and system, which belongs to the field of adaptive optics. The method comprises: (1) selecting a uniformly distributed and representative combination from a certain range of liquid lens structural parameter combinations by means of uniform experimental design; (2) extracting the film force deformation data of the lens under the selected combination during the zooming process by writing a control program; (3) obtaining the optical performance data of the lens under the selected combination during the zooming process by writing a control program; (4) constructing a deep learning model and training the relationship between structural parameters and lens performance; (5) using the model to predict the liquid lens performance data under all structural parameter combinations; (6) setting evaluation criteria in combination with application requirements to screen out the optimal structural parameter combination. The present invention obtains the relationship between the structural parameters and performance of the liquid lens through training through the above optimization method, and finally obtains the optimal liquid lens parameter combination design with excellent dynamic optical performance and low driving force.

Description

Technical Field

[0001] The present invention belongs to the field of adaptive optics, and more specifically, relates to a method and system for optimizing liquid lens performance. Background Art

[0002] Liquid lenses have been widely studied and applied due to their excellent zoom performance achieved with a single lens.

[0003] Chinese patent CN109459851A proposes optimizing the initial focal length and central film thickness to obtain an initial structure for correcting spherical aberration in a liquid lens. This design significantly corrects spherical aberration, but the optimization method fails to analyze all possible structural parameters, resulting in only a locally optimal design, rather than a globally optimal one. Furthermore, the optimization process uses only spherical aberration as an evaluation criterion, ignoring other aberrations (such as coma and field curvature), as well as driving force and response speed.

[0004] The aberration optimization method in the document "Research on Aberration Optimization and Drive Integration of Liquid Lenses with Large Zoom Range" takes into account all geometric aberrations to obtain an optimal non-uniform thickness film structure. However, this optimization method only selects the optimal structure from the optimized surface shapes and film thicknesses at several specific initial focal lengths. However, the parameter combinations of initial surface shapes and film thicknesses are very large, and this optimization method only obtains a locally optimal parameter combination. Outside the parameter selection range of this method, there must be a better global optimal parameter combination. Summary of the Invention

[0005] In response to the shortcomings of the prior art, the purpose of the present invention is to provide a liquid lens performance optimization method based on uniform testing and deep learning, aiming to quickly select a liquid lens non-uniform thickness film structure with excellent optical performance and low driving force from all parameter combinations of the liquid lens through this method.

[0006] To achieve the above objectives, the present invention provides a method for optimizing the performance of a liquid lens, comprising the following steps:

[0007] (1) Using the uniform experimental design method, a portion of uniformly distributed and representative parameter combinations is selected from all parameter combinations of the liquid lens;

[0008] (2) Extraction of simulation data of deformation of liquid lens film: The data of deformation of liquid lens film under the selected parameter combinations are obtained by programming simulation software;

[0009] (3) extracting the performance data of the liquid lens during the zoom process under the above selected parameter combinations by writing a program;

[0010] (4) Constructing a deep learning model with the help of a deep learning framework. The input layer is the liquid lens film structure parameters, and the output layer is the mechanical and optical performance data of the liquid lens under the above-selected parameter combinations. The relationship between the parameter combination and the performance is obtained through training.

[0011] (5) Using the trained deep learning model, predict the performance data under all parameter combinations;

[0012] (6) Setting evaluation criteria and obtaining the liquid lens structure under the optimal parameter combination from all the obtained parameter data.

[0013] Beneficial effect: Through the above-mentioned optimization method, the present invention trains the influence relationship of the parameter combination of the liquid lens structure on its performance, improves the simulation efficiency, and obtains an optimal parameter combination through optimization selection, and finally obtains a liquid lens non-uniform thickness film structure with excellent optical performance, small driving force requirement and short zoom time.

[0014] Preferably, the parameter combination includes: film surface structure parameters and center film thickness parameters.

[0015] Preferably, the performance data includes geometric aberration data, the driving force required for the liquid lens to zoom to the limit state, the zoom response speed, etc.

[0016] Beneficial effect: The selection of the initial structure of the liquid lens not only takes into account the influence of the aberration of the focal length adjustment range, but also takes into account the requirements of the driving force and the response speed, so that the optimized optimal structure not only has excellent optical performance but is also easy to drive.

[0017] Preferably, the evaluation criteria are: first normalize the geometric aberrations, driving force data, response speed, etc., and reasonably set the weights of the geometric aberrations, driving force data, response speed, etc. according to actual needs, and calculate the weighted sum of the normalized geometric aberrations, driving force data and response speed, etc. as the evaluation criteria for its performance.

[0018] Beneficial effects: For different application scenarios, the optimal structure suitable for different scenarios can be obtained by changing the weight setting, making the optimization method have excellent universality.

[0019] The present invention also provides a liquid lens performance optimization system, comprising: a computer-readable storage medium and a processor;

[0020] The computer-readable storage medium is used to store executable instructions;

[0021] The processor is used to read the executable instructions stored in the computer-readable storage medium and execute the above-mentioned liquid lens performance optimization method.

[0022] Compared with the prior art, the above technical solutions conceived by the present invention can achieve the following beneficial effects.

[0023] 1. The liquid lens performance optimization method based on uniform testing and deep learning provided by the present invention obtains the influence of the parameter combination of the liquid lens structure on its performance by constructing a deep learning model for training, and then can quickly select a globally optimal parameter combination from all parameter combinations.

[0024] 2. The liquid lens performance optimization method based on uniform testing and deep learning provided by the present invention selects the initial structure of the liquid lens not only considering the influence of aberrations within the focal length adjustment range, but also considering the requirements of driving force, so that the optimized optimal structure not only has excellent optical performance but is also easy to drive.

[0025] 3. The liquid lens performance optimization method based on uniform testing and deep learning provided by the present invention selects the initial structure of the liquid lens not only considering the influence of aberrations within the focal length adjustment range, but also considering the requirements of response speed, so that the optimized optimal structure not only has excellent optical performance but also can adapt to the needs of practical applications.

[0026] 4. The electromagnetically driven bidirectional variable focus liquid lens provided by the present invention can obtain the optimal structure suitable for different application scenarios by changing the weight setting, making the optimization method have excellent universality. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 A schematic structural diagram of the liquid lens provided by the present invention;

[0028] Figure 2 A schematic flow chart of the liquid lens performance optimization method based on uniformity testing and deep learning provided by the present invention;

[0029] Figure 3 Schematic diagram of the constructed deep learning model;

[0030] Figure 4 The training effect of the constructed deep learning model on spherical aberration;

[0031] Figure 5 The training effect of the constructed deep learning model on coma;

[0032] Figure 6 The training effect of the deep learning model built for driving force;

[0033] Figure 7 The spherical aberration data under all parameter combinations predicted by the trained deep learning model;

[0034] Figure 8The coma data under all parameter combinations predicted by the trained deep learning model;

[0035] Figure 9 The driving force data under all parameter combinations predicted by the trained deep learning model;

[0036] Figure 10 Comparison of spherical aberration curves for four structures;

[0037] Figure 11 Comparison of coma curves of four structures. DETAILED DESCRIPTION

[0038] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0039] The performance of the liquid lens is mainly related to the parameter values ​​of the non-uniform thickness film structure of the liquid lens, such as Figure 1 As shown in the figure, the parameters of the structure include three even-order aspheric surface parameters a, b, c and the center film thickness parameter h, a total of four parameters. In order to obtain the best parameter values, it is necessary to obtain the performance data of the liquid lens under different parameter combinations. If each parameter has n values, there will be n 4 Even if each parameter has only 1000 possible values, the total number of parameter combinations will be as high as 1000. 4 It is extremely difficult and time-consuming to obtain data under so many parameter combinations. Therefore, the present invention develops a liquid lens performance optimization method based on uniform experiment and deep learning. Taking n as 1000 as an example, Figure 2 As shown, it includes the following steps:

[0040] (1) Using the uniform experimental design method, a number of evenly distributed and representative parameter combinations are selected from all parameter combinations of the liquid lens, as shown in Table 1;

[0041] Table 1

[0042] Parameter combination index value Central membrane thickness h Quadratic term coefficient a The coefficient of the fourth term b Sixth-order coefficient c 1 0.2005 0.01440 -5.600e-06 -1.568e-05 2 0.2010 0.02785 8.840e-06 -1.132e-05 3 0.2015 0.04130 -1.672e-05 -6.960e-06 4 0.2020 0.00475 -2.280e-06 -2.600e-06 … … … … … 999 0.6995 0.03750 5.520e-06 1.560e-05 1000 0.7000 0.05095 1.996e-05 1.996e-05

[0043] (2) Extraction of simulation data of force deformation of liquid lens film: The Comsol simulation is controlled by Matlab program to obtain the force deformation data of the liquid lens film under the above selected parameter combinations;

[0044] (3) Using Matlab to control Zemax, the performance data of the liquid lens during the zoom process under the selected parameter combinations were extracted;

[0045] (4) Build deep learning models with the help of deep learning frameworks, such as Figure 3 As shown, the input layer is the selected parameter combination, and the output layer is the performance data of the liquid lens under the selected parameter combination. The relationship between the parameter combination and the performance is obtained through training, as shown in Figure 4 Shown is the training effect of spherical aberration, Figure 5 is the training effect of coma, Figure 6 The training effect of the driving force is shown in Figure 1. In the figure, "o" represents the actual value of the performance data obtained according to the above simulation process, and "+" represents the predicted value of the performance data predicted by the trained deep learning model. It can be seen from the figure that the two are highly consistent, which further demonstrates that the training effect of the deep learning model is very good.

[0046] (5) Using the trained deep learning model, predict the performance data under all parameter combinations, such as Figure 7 Shown are the spherical aberration data under all parameter combinations predicted by the trained deep learning model. Figure 8 is the predicted coma data, Figure 9 is the predicted driving force data;

[0047] (6) Set evaluation criteria, analyze the performance data under all predicted parameter combinations, and select the best parameter combination.

[0048] In order to verify the beneficial effects of the optimization method provided by the present invention, the performance of the liquid lens under the optimal parameter combination obtained by optimization was compared with the other three structures, such as Figure 10 The following is a comparison of the spherical aberration curves of the four structures. Figure 11 Table 2 is a coma curve comparison chart showing the driving force required for the four liquid lenses to zoom to the extreme state. The 0.1 mm uniform thickness film structure is a structure commonly used in conventional liquid lens designs, the 0.15 mm uniform thickness film structure is a structure with a driving force close to that required for the non-uniform thickness film optimized by the present invention, the non-uniform thickness film structure 1 is the structure optimized in the document "Study on Aberration Optimization and Drive Integration of Liquid Lenses with a Large Zoom Range", and the non-uniform thickness film structure 2 is the structure optimized by the present invention. It can be seen that the non-uniform thickness film structure optimized by the present invention has greatly improved performance in spherical aberration, coma, and driving force compared to the other three structures.

[0049] Table 2

[0050] Thin film structure Driving force / N 0.1mm thick film structure 0.0570 0.15mm thick film structure 0.0729 Non-uniform thickness film structure 1 0.1470 Non-uniform thickness film structure 2 0.0729

[0051] It will be easily understood by those skilled in the art that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for optimizing liquid lens performance, characterized in that: The following steps are involved: (1) Using a uniform experimental design method, a portion of uniformly distributed parameter combinations is selected from all structural parameter combinations of the liquid lens within a preset range, wherein the parameter combinations include film surface structural parameters and thickness parameters; (2) Extraction of simulation data of force deformation of liquid lens film: Analyze and extract the force deformation data of the liquid lens film during the zoom process under the above selected parameter combinations; (3) Extraction of simulation data of optical performance of liquid lens: Analyze and extract the optical performance data of liquid lens during zooming under the selected parameter combinations; (4) Using a deep learning framework to construct a deep learning model, the input layer is the combination of liquid lens film parameters, and the output layer is the liquid lens performance data under the corresponding structural parameter combination, and the relationship between the structural parameters and the lens performance is obtained through training; (5) Using the trained deep learning model, traverse and predict the performance data of the liquid lens under all parameter combinations; (6) According to the application requirements, the corresponding evaluation criteria are set to obtain the optimal liquid lens structure from all the data. The evaluation criteria are as follows: first, the geometric aberration and driving force are normalized, and the weights of each item are reasonably set according to actual needs. The weighted sum of the normalized geometric aberration, driving force, and is calculated as the comprehensive performance evaluation index of the liquid lens.

2. The liquid lens performance optimization method according to claim 1, characterized in that: In the step (2), the film deformation data includes the driving force required when the liquid lens is zoomed to the limit state.

3. The liquid lens performance optimization method according to claim 1, characterized in that: In the step (3), the optical performance data includes geometric aberrations.

4. The liquid lens performance optimization method according to claim 2, characterized in that: The simulation data of the stress and deformation of the liquid lens film are obtained by controlling the Comsol simulation with Matlab program.

5. The liquid lens performance optimization method according to claim 3, characterized in that: The optical performance simulation data of the liquid lens is extracted by controlling Zemax using Matlab.

6. A liquid lens performance optimization system, characterized in that: include: Computer-readable storage media and processor; The computer-readable storage medium is used to store executable instructions; The processor is configured to read the executable instructions stored in the computer-readable storage medium and execute the liquid lens performance optimization method according to any one of claims 1 to 5.

Images