Wavefront aberration prediction method for optomechanical system under variable temperature conditions

Abstract

The present invention discloses a method for predicting the wavefront aberration of an optical-mechanical system under variable temperature conditions, comprising the following steps: step 1, establishing a physical model and a simulation model of an optical system; step 2, measuring and calibrating the simulation model with the physical model of the optical system; step 3, constructing a neural network data set with the simulation model; step 4, training the neural network model: the convolutional neural network comprises an input layer, a convolutional layer, a pooling layer, a fully connected layer and an output layer; the convolutional neural network is trained using the samples obtained in step 3, and when the loss function of the convolutional neural network converges, the entire training process is completed; step 5, predicting the wavefront aberration of the system using the convolutional neural network. The present invention uses a temperature sensor to monitor the temperature of key points in real time, and by training a deep learning network, the surface performance of the optical-mechanical system is predicted in real time, which can effectively improve the error compensation efficiency and accuracy of the system.

Description

Technical Field

[0001] The present invention belongs to the field of optics, and in particular relates to a wavefront aberration prediction method for an optomechanical system under variable temperature conditions, and can be applied to the prediction and evaluation of thermal error deformation of an optomechanical system caused by temperature changes. Background Art

[0002] With the application of laser technology in the field of equipment, the wavefront aberration accuracy requirements of optical machine output are getting higher and higher. However, under variable temperature conditions such as aerospace, the overall structure of different materials in the optical system has uneven deformation under temperature changes, which in turn causes changes in optical imaging accuracy, making it difficult to meet high-precision application scenarios.

[0003] Traditionally, the adjustment method for optomechanical thermal errors mainly uses active adjustment methods. The active adjustment method uses a wavefront sensor to detect wavefront changes and errors in real time, calculates the correction amount in real time through a computer control system, drives the corrector to output a certain force, torque or displacement, and corrects the system wavefront error in real time to ensure that the optical imaging quality meets the requirements. However, the active adjustment method relies on the wavefront sensor to measure the real-time aberration. The wavefront sensor has a complex structure and heavy weight, and is difficult to adapt to the optical system in the aerospace field with small space and high lightweight requirements. Therefore, finding a more convenient method to measure wavefront aberrations is important for high-precision optomechanical systems under variable temperature conditions. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a method for predicting the wavefront aberration of the optomechanical system under variable temperature conditions by using a temperature sensor to monitor the temperature of key points in real time and by training a deep learning network; after the model parameters of the neural network are trained, a method for predicting the wavefront aberration of the optomechanical system under variable temperature conditions can be used to effectively improve the error compensation efficiency and accuracy of the system.

[0005] The objective of the present invention is achieved through the following technical solution: A method for predicting the wavefront aberration of an optical mechanical system under variable temperature conditions, characterized in that it comprises the following steps:

[0006] Step 1. Establish the physical model and simulation model of the optical system: Complete the three-dimensional structural design of the optical system, and process and assemble the physical model of the optical-mechanical system. Import the established three-dimensional model into the Themal-Structural module of the Ansys software, load the ambient temperature as the boundary condition, and establish the simulation model;

[0007] Step 2: Use the physical model of the optical system to measure and calibrate the simulation model, including the following steps:

[0008] Step 2.1, selection of typical temperature points and temperature simulation calibration: In Ansys software, boundary conditions are set according to the actual ambient temperature, and the overall temperature of the optical-mechanical system is simulated and calculated, and n points with large temperature differences are found as typical temperature points, n ≥ 10; temperature sensors are attached to the corresponding points of the physical model, and the actual temperature of the points is measured. The simulation model is modified according to the actual data until the temperature difference between the temperature test data and the simulation data is less than 0.1°C;

[0009] Step 2.2, surface shape measurement and calibration under typical temperature conditions: affix a high-precision temperature sensor to a typical temperature point of the actual optical system, select multiple typical temperature conditions in the entire variable temperature ambient temperature, collect the temperature sensor data under the ambient temperature, and measure the corresponding optical system surface shape data; perform simulation calculation on the surface shape under the temperature condition, compare the actual surface shape value with the simulated surface shape value, and adjust the simulation settings accordingly until the error between the actual surface shape value and the simulated surface shape value is less than 1%;

[0010] Step 3: Use the simulation model to build a neural network data set, including the following steps:

[0011] Step 3.1, within the entire temperature variation range, select N ambient temperature point data as sample points according to the equal temperature difference, and calculate the temperature and displacement in the Ansys simulation model;

[0012] Step 3.2, record the simulated temperature data at the typical temperature points, derive the mirror node displacement data, calculate the mirror rigid body displacement and the Zernike coefficient, and use the temperature data at the typical temperature points, the mirror rigid body displacement and the Zernike coefficient as a set of samples;

[0013] Step 3.3, repeat step 3.2 until N groups of samples are obtained;

[0014] Step 4: Train the neural network model: Select a convolutional neural network suitable for image regression problems. The convolutional neural network includes an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an output layer. Use the samples obtained in step 3 to train the convolutional neural network. When the loss function of the convolutional neural network converges, the entire training process is completed.

[0015] Step 5: Predict wavefront aberration: Input n temperature points into the network to measure the temperature, and use the convolutional neural network obtained in step 4 to predict the system wavefront aberration under the temperature condition.

[0016] The beneficial effects of the present invention are:

[0017] 1. The traditional wavefront sensor has a complex structure and is heavy. The present invention only needs to use a few sensors to collect data, and then predict the aberration through a neural network, which can effectively reduce the complexity of the overall system and meet higher lightweight requirements.

[0018] 2. Use temperature sensors to monitor the temperature of key points in real time, and train deep learning networks to achieve a nonlinear mapping relationship between key point temperatures and optical mirror surface shapes, and make real-time predictions of the surface shape performance of the optical-mechanical system. After the model parameter training of the neural network is completed, the input of key temperature point data can accurately output the predicted surface shape value, which can effectively improve the error compensation efficiency and accuracy of the system.

[0019] 3. Neural networks have strong universality. When facing a new thermal error optical system using the idea of ​​transfer learning, a small sample is established for training based on the previously trained neural network model parameters, which can greatly reduce the training time. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 The present invention is a flow chart of the method for predicting the wavefront aberration of an optical-mechanical system under variable temperature conditions. DETAILED DESCRIPTION

[0021] This embodiment is described in detail by taking an optical system under working conditions of -20°C to 60°C as an example, and the technical solution of the present invention is further described below with reference to the accompanying drawings.

[0022] like Figure 1 As shown, a method for predicting a wavefront aberration of an optical mechanical system under variable temperature conditions of the present invention comprises the following steps:

[0023] Step 1. Establish the physical model and simulation model of the optical system: Complete the three-dimensional structural design of the optical system, and process and assemble the physical model of the optical-mechanical system. Import the established three-dimensional model into the Themal-Structural module of the Ansys software, load the ambient temperature as the boundary condition, and establish the simulation model;

[0024] Step 2: Use the physical model of the optical system to measure and calibrate the simulation model, including the following steps:

[0025] Step 2.1, selection of typical temperature points and temperature simulation calibration: In Ansys software, boundary conditions are set according to the actual ambient temperature, and the overall temperature of the optical-mechanical system is simulated and calculated, and n points with large temperature differences are found as typical temperature points, n ≥ 10; temperature sensors are attached to the corresponding points of the physical model, and the actual temperature values ​​of the points are measured. The material parameters, contact conditions and other parameters in the simulation model are modified according to the actual data until the temperature difference between the temperature test data and the simulation data is less than 0.1°C;

[0026] Step 2.2, typical temperature condition surface shape measurement and calibration: a high-precision temperature sensor is attached to a typical temperature point of the actual optical system, and multiple typical temperature conditions are selected in the entire variable temperature condition ambient temperature. In this embodiment, 2 groups of typical temperature conditions are selected in each of the three seasons of spring, summer and winter, and the temperature sensor data under the ambient temperature is collected, and the corresponding optical system surface shape data is measured; and the surface shape under the temperature condition is simulated and calculated, and the actual surface shape value is compared with the simulated surface shape value, and the simulation settings are adjusted accordingly until the error between the actual surface shape value and the simulated surface shape value is less than 1%;

[0027] Step 3: Use the simulation model to build a neural network data set, including the following steps:

[0028] Step 3.1, within the entire temperature variation range, select N ambient temperature point data as sample points according to the equal temperature difference, and calculate the temperature and displacement in the Ansys simulation model; in this embodiment, within the entire temperature variation range (-20°C to 60°C), select 160 ambient temperature point data as sample points according to the equal temperature difference (0.5°C);

[0029] Step 3.2, record the simulated temperature data at the typical temperature points, derive the mirror node displacement data, calculate the mirror rigid body displacement and the Zernike coefficient, and use the temperature data at the typical temperature points, the mirror rigid body displacement and the Zernike coefficient as a set of samples;

[0030] Step 3.3, repeat step 3.2 until 160 groups of samples are obtained, that is, a complete sample set is obtained. The sample set can be divided into a training set and a test set. The training set is used to train the convolutional neural network, and the test set is used to verify the fitting ability of the neural network;

[0031] Step 4: Train the neural network model: Select a convolutional neural network suitable for image regression problems. The convolutional neural network includes an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an output layer. Use the samples obtained in step 3 to train the convolutional neural network. When the loss function of the convolutional neural network converges, the entire training process is completed.

[0032] Step 5: Predict wavefront aberration: Input n temperature points into the network to measure the temperature, and use the convolutional neural network obtained in step 4 to predict the system wavefront aberration under the temperature condition.

[0033] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.

Claims

1. A method for predicting the wavefront aberration of an optomechanical system under variable temperature conditions, characterized in that: The following steps are involved: Step 1: Establish the physical model and simulation model of the optical system: complete the three-dimensional structural design of the optical system, and process and assemble the physical model of the optical-mechanical system; import the established three-dimensional model into the Themal-Structural module of Ansys software, load the ambient temperature as the boundary condition, and establish the simulation model; Step 2: Use the physical model of the optical system to measure and calibrate the simulation model, including the following steps: Step 2.1, selection of typical temperature points and temperature simulation calibration: In Ansys software, boundary conditions are set according to the actual ambient temperature, and the overall temperature of the optical-mechanical system is simulated and calculated, and n points with large temperature differences are found as typical temperature points, n ≥ 10; temperature sensors are attached to the corresponding points of the physical model, and the actual temperature of the points is measured. The simulation model is modified according to the actual data until the temperature difference between the temperature test data and the simulation data is less than 0.1°C; Step 2.2, surface shape measurement and calibration under typical temperature conditions: affix a high-precision temperature sensor to a typical temperature point of the actual optical system, select multiple typical temperature conditions in the entire variable temperature ambient temperature, collect the temperature sensor data under the ambient temperature, and measure the corresponding optical system surface shape data; perform simulation calculation on the surface shape under the temperature condition, compare the actual surface shape value with the simulated surface shape value, and adjust the simulation settings accordingly until the error between the actual surface shape value and the simulated surface shape value is less than 1%; Step 3: Use the simulation model to build a neural network data set, including the following steps: Step 3.1, within the entire temperature variation range, select N ambient temperature point data as sample points according to the equal temperature difference, and calculate the temperature and displacement in the Ansys simulation model; Step 3.2, record the simulated temperature data at the typical temperature points, derive the mirror node displacement data, calculate the mirror rigid body displacement and the Zernike coefficient, and use the temperature data at the typical temperature points, the mirror rigid body displacement and the Zernike coefficient as a set of samples; Step 3.3, repeat step 3.2 until N groups of samples are obtained; Step 4: Train the neural network model: Select a convolutional neural network suitable for image regression problems. The convolutional neural network includes an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an output layer. Use the samples obtained in step 3 to train the convolutional neural network. When the loss function of the convolutional neural network converges, the entire training process is completed. Step 5: Predict wavefront aberration: Input n temperature points into the network to measure the temperature, and use the convolutional neural network obtained in step 4 to predict the system wavefront aberration under the temperature condition.

Images