Active optical correction closed-loop solving method based on influence function iterative updating

Abstract

The invention relates to an active optical correction closed-loop solving method based on influence function iterative updating. The method comprises the following steps: acquiring an initial influence function of each actuating unit and carrying out abnormal point replacement; the wavefront sensing system detects the wavefront aberration of the system, calculates the voltage to be applied to each actuating unit according to the wavefront aberration of the system in an ideal state and an initial standard influence function relationship, and applies the voltage to the corresponding actuating unit; the wavefront aberration of the system is detected again, the offset coefficient of each influence function is calculated according to the difference between the control surface shape residual error in the actual state and the control surface shape residual error in the ideal state, each influence function is updated according to the offset coefficient, and a new influence function is obtained; and taking the new influence function as an initial standard influence function in the next round of regulation and control to realize the correction value closed-loop solution of the wavefront corrector. According to the method, the real-time iteration updating of the influence function is realized, and the calculation accuracy of the active optical correction quantity is improved.

Description

Technical Field

[0001] The invention relates to the technical field of optical system aberration correction, and in particular to a closed-loop solution method for active optical correction amount based on iterative updating of influence functions. Background Art

[0002] As the requirements for the resolution and light-gathering capabilities of space telescopes continue to increase, the aperture of the main reflector of the optical system is getting larger and larger. Compared with small-aperture reflectors, large-aperture reflector assemblies are more sensitive to environmental interference such as gravity changes, temperature changes, transportation and launch, and are more difficult to manufacture and assemble. This makes it difficult to ensure the surface accuracy of the main reflector, which in turn causes a decrease in the imaging performance of the optical system. How to ensure the high-performance application of space optical systems on orbit is an important issue in the development of large-aperture space optical systems, and maintaining the surface accuracy of the main reflector is a key link.

[0003] Active optical technology is achieved by adding an actuator unit to the back of the mirror body, which can achieve nanometer-level surface accuracy adjustment of the mirror, thereby compensating for manufacturing and adjustment residuals and environmental errors.

[0004] Active optical systems generally consist of three components: a wavefront sensor, a wavefront controller, and a wavefront corrector. The working process is as follows: the wavefront sensor is used to measure the aberration of the incident wavefront, the wavefront controller converts the aberration into a corresponding control signal and transmits it to the wavefront corrector. After the wavefront corrector receives the signal, it controls the actuator output to apply a correction force to the mirror surface to correct the wavefront aberration. Therefore, the wavefront corrector is the core component of the active optical system. The accuracy of the correction or adjustment amount of the wavefront corrector directly affects the accuracy of the target aberration correction. The schematic diagram of the active deformable mirror structure is shown in the figure. Figure 1 As shown, Figure 1 The deformable mirror structure shown is only one of many structures, and other deformable mirror structures may also be used.

[0005] There are two main algorithms for calculating the correction or adjustment amount of the wavefront corrector:

[0006] Algorithm 1: Single open-loop solution algorithm:

[0007] The main principle of the algorithm is as follows: Gaussian function is used to simulate and generate influence function, and open-loop solution algorithm is used to solve the voltage value of the actuator. The principle of the correction amount solution algorithm is: after measuring the influence function of each actuator, we can get the correction target aberration W(x, y) and the voltage v that should be applied to each actuator. n The relationship between.

[0008]

[0009] Among them, I n(x, y) is the influence function of the nth actuator, N is the number of actuators;

[0010] If the number of sampling points of the deformable mirror shape is M, the coordinates of the sampling points are (x m ,y m ), m = 1, 2, ..., M, the wavefront aberration value at the mth point is W(x m ,y m ), then the matrix equation shown in formula (2) can be listed, and the voltage matrix V can be obtained by using the least square method. This process is an open-loop process, and its influence function matrix is ​​measured one by one after the mirror components are installed. In the control process after a measurement is completed, the influence function is no longer updated.

[0011]

[0012] Nian Wei, Liu Zhaojun and others from the Beijing Institute of Space Mechanics and Electronics used this algorithm to analyze the correction capabilities of deformable mirrors of large-aperture space telescopes.

[0013] The adjustment amount calculation algorithm of Algorithm 1 mainly includes three steps: measuring the influence function, measuring the wavefront aberration, and calculating the adjustment amount.

[0014] Step 1: Measure the influence function of each actuator. After the deformable mirror is processed and assembled, the influence function of each actuator needs to be calibrated and measured. The measurement steps are as follows: ① The deformable mirror and the interferometer form a measurement optical path, and record the state of the deformable mirror surface under the reference state; ② Apply a unit voltage to each actuator unit separately, and measure the surface shape change caused by the unit control voltage. The measurement result is the influence function I of this actuator unit. n (x, y); ③ The influence functions of all the actuator units measured are combined into an influence function matrix and stored as known parameters;

[0015] Step 2: Under the actual working state of the test, the wavefront aberration of the deformable mirror is recorded as W(x, y) for measurement, and the aberration information is transmitted to the wavefront controller component;

[0016] Step 3: Calculate the voltage v of each actuator unit through the correction amount solution algorithm n , and then the wavefront corrector controls the surface shape to produce corresponding deformation, thereby achieving the purpose of offsetting the aberration.

[0017] In Algorithm 1, the linear superposition relationship between the wavefront aberration and the influence function is adopted, and the voltage to be applied to each actuator is solved by the least square method. However, Algorithm 1 has the following four main disadvantages:

[0018] 1. In the existing methods, the influence function of the actuator is usually calibrated once and reused multiple times. However, in the actual working process, the influence function of the actuator will cause offset changes due to external disturbances and return errors. Therefore, continuing to use the previously measured influence function to calculate the adjustment amount is no longer accurate, which leads to the inability to accurately compensate for the aberration;

[0019] 2. When the influence function is offset in actual use, the influence function needs to be recalibrated. Since the traditional calibration method requires the actuator influence function to be measured and calibrated one by one, the calibration process will be complicated and time-consuming, and the calibration time will increase linearly with the increase in the number of actuators;

[0020] 3. In the existing calibration and measurement process of the actuator influence function, the measured influence function is not subjected to secondary processing. The influence function contains burrs and abnormal points caused by the internal noise and resolution of the interferometer, resulting in reduced solution accuracy.

[0021] 4. The process of calculating the actuator voltage from the target surface shape is a high-dimensional (multi-variable) optimization problem. The algorithm generally uses the least squares method. In this solution model, the adjustment voltage of each actuator is an optimization variable. In the current calculation, only one round of calculation is performed, and the result obtained is not the optimal solution.

[0022] Algorithm 2: An adaptive optical deformable mirror control method based on attention mechanism disclosed in the invention patent with publication number CN117970799A;

[0023] This algorithm was proposed by the Institute of Optoelectronics Technology of the Chinese Academy of Sciences. Through open-loop control in a simulation environment, high-performance control commands are obtained as a data set for reinforcement learning. After transfer learning, the reinforcement learning model can be quickly migrated to different turbulent environments and the robustness of model control can be improved. Compared with traditional wavefront correction, this method can overcome the problem that traditional proportional integral control methods cannot effectively cope with turbulent changes. The main process of the algorithm is:

[0024] Construct a reinforcement learning model based on the actor-critic algorithm; collect the optimal control command sequence and the corresponding PSF image sequence of adaptive optical open-loop control in a simulation environment, and put them into the experience pool; train the reinforcement learning model based on the obtained image sequence and the optimal control command sequence; collect image sequence information of the real-time optical environment and calculate the control command; control the deformable mirror according to the calculated control command; put the obtained control command and PSF image sequence information into the experience pool; update the reinforcement learning model based on the control sequence data set put into the experience pool.

[0025] Algorithm 2 uses a pre-trained reinforcement learning model to solve the control voltage, but Algorithm 2 has the following two main disadvantages:

[0026] 1. This algorithm is mainly used in the field of adaptive optics to combat atmospheric turbulence, but is not completely applicable to active optics in the field of space optics;

[0027] 2. Existing deep learning methods require a large amount of data as training sets, and model training takes a long time. In addition, deep learning models are black box models with no mathematical basis and lack of practical application reliability. Summary of the invention

[0028] In response to the problems existing in the existing wavefront corrector correction amount calculation algorithm, the present invention proposes a closed-loop solution method for active optical correction amounts based on iterative updating of influence functions. This method improves the problems existing in the algorithm in the prior art by correcting the residual solution actuator influence function deviation and calibrating it in real time, thereby realizing real-time iterative updating of the influence function and further realizing accurate calculation of the active optical correction amount.

[0029] To solve the above problems, the present invention adopts the following technical solutions:

[0030] A closed-loop solution method for active optical correction based on iterative update of influence functions, the method comprising the following steps:

[0031] Step 1: Collect the initial influence function I0 of each actuator unit in the current wavefront corrector n (x, y), and the initial influence function I0 n (x, y) is used to replace the abnormal point and obtain the initial standard influence function I of each actuator unit n (x, y), where n is the number of the actuator unit, n = 0, 1, 2, ..., N;

[0032] Step 2: Use the wavefront sensing system to detect the system wavefront aberration W1(x, y) before correction, and according to the system wavefront aberration W1(x, y) in the ideal state and the initial standard influence function I of each actuator unit n (x, y) and the voltage v applied to each actuator unit n The matrix equation between is solved to obtain the voltage v1 that should be applied to each actuator unit. n , and then obtain the control surface shape and control surface shape residual ΔW in the ideal state respectively ideal (x,y), and according to the calculated voltage v1 n It is applied to the corresponding actuating unit to control the mirror surface to undergo corresponding deformation;

[0033] Step 3: The wavefront sensing system detects the system wavefront aberration again to obtain the control surface shape and the control surface shape residual ΔW in the actual state. real (x, y), and the control surface shape residual ΔW in the actual statereal The residual error ΔW between (x, y) and the ideal control surface shape ideal (x, y) is subtracted to obtain ΔW p (x,y), the formula is as follows:

[0034] ΔW p (x,y)=ΔW real (x,y)-ΔW ideal (x,y) (13)

[0035] Substitute the calculation expression of each part of the control surface residual into it and get:

[0036]

[0037] Among them, k n is the offset coefficient of the nth influence function;

[0038] Step 4: According to ΔW shown in formula (14) p The relationship between (x, y) and the influence function used in this round of regulation is solved to calculate the offset coefficient of each influence function, and each influence function is updated once according to the offset coefficient to obtain a new influence function, and then return to step 2, and use the new influence function as the initial standard influence function in the next round of regulation, and repeat steps 2 to 4 to realize the closed-loop solution of the correction amount of the wavefront corrector.

[0039] The closed-loop solution method for active optical correction based on iterative update of influence function proposed in the present invention has the following advantages:

[0040] (1) To address the problem that the influence function of the actuator unit may be offset during use, the present invention adopts an iteratively updated influence function. After each control process, the influence function is iteratively updated in real time according to the control residual to ensure the accuracy of the influence function, thereby improving the accuracy of the calculation of the active optical correction amount;

[0041] (2) In view of the problem that the measurement and calibration methods of traditional influence functions are complicated and time-consuming, the present invention reversely calculates the influence function offset by the difference between the actual control residual and the ideal control residual, thereby achieving the effect of calculating all influence function offsets once, thus avoiding the time consumption caused by the need to recalibrate the influence function during use;

[0042] (3) The present invention adopts a method for obtaining the gradient of the influence function, screens out points where the gradient produces a sudden change, and fills the abnormal points with the closest non-outlier value, thereby reducing the error introduced in the actual measurement of the influence function;

[0043] (4) For the problem that the solution result is not the optimal solution, the present invention adopts a secondary solution strategy to achieve the optimal solution based on the relationship curve between the number of solutions and the solution accuracy;

[0044] (5) The present invention takes into account the law of environmental disturbances in space optical systems, and has a good correction effect on the deviation of the influence function caused by regular temporal low-frequency disturbances, such as orbital periodicity, radiation periodicity, etc.;

[0045] (6) Compared with deep learning methods, the model does not require a large amount of data set training. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without exceeding the scope of protection required by the present invention.

[0047] Figure 1 This is a schematic diagram of the construction principle of the active deformable mirror;

[0048] Figure 2 It is a flow chart of a closed-loop solution method for active optical correction value based on iterative update of influence function according to an embodiment of the present invention;

[0049] Figure 3 The nonlinear “displacement-voltage” response curve of the piezoelectric ceramic of the actuator unit;

[0050] Figure 4 Comparison chart of the correction residual RMSE after multiple rounds of aberration correction under different conditions. DETAILED DESCRIPTION

[0051] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0052] In one embodiment, if Figure 2 As shown, the present invention provides a closed-loop solution method for active optical correction value based on iterative update of influence function, and the method includes the following steps 1 to 4.

[0053] Step 1: Collection and processing of the initial influence function. In this step, a unit voltage is first applied to each actuator unit in the wavefront corrector (taking the deformable mirror as an example), and the initial influence function of each actuator unit is measured by the interferometer; then the gradient of the initial influence function is solved, and the coordinates of the mutation points in the gradient map are marked; then the values ​​at the coordinates of the mutation points in the influence function map are replaced by the nearest point interpolation to eliminate the abnormal values ​​of the initial influence function caused by the interferometer, and finally the initial standard influence function of each actuator unit is obtained.

[0054] Specifically, assuming that there are N actuator units in the deformable mirror, the initial influence function of each actuator unit is obtained by interferometer acquisition and measurement. The initial influence function can be expressed as I0 n (x, y), where n is the number of the actuator unit, I0 n (x, y) represents the initial influence function of the nth actuator unit. Next, the initial influence function I0 n (x, y) is used to replace outliers.

[0055] Step 1.1: During the acquisition of the initial influence function, due to factors such as the internal noise of the interferometer, burrs will appear at the edge of the detection, resulting in the appearance of abnormal values ​​in the initial influence function. If it is not processed, it will affect the solution result. Therefore, the initial influence function I0 is calculated first. n The image gradient information of (x, y) is calculated as follows:

[0056] (Gx, Gy) = Gradient (I0 n (x,y)) (3)

[0057] Among them, Gx and Gy are the gradients in the x direction and y direction respectively.

[0058] Step 1.2: Identify the mutation point in the gradient map as an outlier and record the coordinates of the outlier (x0, y0), as shown in formula (4). Find the coordinates of the outlier by using the find() function, and then use the nearest point interpolation to replace the value at the outlier, as shown in formula (5):

[0059]

[0060] I0 n (x0,y0)=Nearest(x0,y0) (5)

[0061] Among them, T0 is the outlier threshold, which is generally defined as the value that differs from the mean by more than three standard deviations. Nearest(x0,y0) is the nearest point interpolation function, which uses the closest non-outlier value to fill the outlier point, that is, the value at the coordinate (x0,y0) in the initial influence function diagram is replaced by the nearest point interpolation to eliminate the outliers brought by the interferometer in the initial influence function.

[0062] Step 1.3: After all abnormal points are replaced, the initial standard influence function I of the actuator unit is obtained. n (x, y).

[0063] Step 2: The wavefront sensing system measures the system wave aberration for the first correction. In this step, the wavefront sensing system is used to detect the system wavefront aberration. The wavefront sensing system mentioned in the present invention does not only refer to actual devices, but also refers to any physical device that can obtain the system wavefront information, including but not limited to wavefront sensors, or image processing algorithms, etc.

[0064] After the first system wavefront aberration W1(x, y) to be corrected is measured by the wavefront sensing system, the system wavefront aberration W1(x, y) and the initial standard influence function I of each actuator unit are n (x, y) and the voltage v applied to each actuator unit n The relationship between them is listed, and the control voltage matrix V of this round of regulation is obtained by solving the relationship between the system wavefront aberration and the influence function.

[0065] In an ideal state (i.e., when the influence function is not shifted), the system wavefront aberration W1(x, t) and the initial standard influence function I n (x, y) and voltage v n The relationship between them is shown as follows:

[0066]

[0067] If the number of sampling points of the deformable mirror shape is M, the coordinates of the sampling points are (x m ,y m ), m = 1, 2, ..., M, the wavefront aberration value at the mth point is W1(x m ,y m ), then according to formula (6), the matrix equation shown in formula (7) can be listed, and the matrix equation is solved using the twice least square method to obtain the voltage v1 that should be applied to each actuator unit: n , forming a control voltage matrix V. Optionally, the least squares method used in this step can also be replaced by other optimization algorithms, including but not limited to genetic algorithms, simulated annealing algorithms, etc.

[0068]

[0069] The ideal control surface shape is:

[0070]

[0071] The control surface shape residual under ideal conditions (i.e., ideal correction residual) is:

[0072] ΔW ideal (x,y)=W ideal (x,y)-W1(x,y) (9)

[0073] Calculate the voltage v1 that should be applied to each actuator unit n After that, the wavefront control system will solve the various voltages v1 n It is applied to the corresponding actuating unit on the wavefront corrector to control the mirror surface of the deformable mirror to undergo corresponding deformation, thereby achieving surface correction.

[0074] Step 3: Back-calculate the influence function deviation based on the control results.

[0075] In actual working conditions, the influence function is not stable and will be offset due to the return error of the actuator unit and environmental factors. Suppose the influence function after offset is k n I n (x,y),k n is the offset coefficient of the nth influence function. Since the offset is unknown when calculating the voltage, the initial standard influence function under the ideal state is still substituted into the calculation. The voltage applied to each actuator unit is solved to be v n However, the control surface shape generated in the actual state is:

[0076]

[0077] The control surface shape residual in the actual state (i.e. the actual correction residual) is:

[0078] ΔW real (x,y)=W real (x,y)-W1(x,y) (11)

[0079] After transforming formula (11), we can get:

[0080] ΔW real (x,y)=W real (x,y)-W ideal (x,y)+W ideal (x,y)-W1(x,y)=ΔW p (x,y)+ΔW ideal (x,y) (12)

[0081] Defining ΔW p (x,y)=W real (x,y)-W ideal (x, y) is the difference between the ideal control surface and the actual control surface. According to formula (12), we have:

[0082] ΔW p (x,y)=ΔW real (x,y)-ΔW ideal (x,y) (13)

[0083] Substituting the calculation expression of each part of the control surface residual, we can get ΔW p The relationship between (x, y) and the influence function used in this round of regulation is shown in the following formula:

[0084]

[0085] Formula (14) is expressed in matrix form as follows:

[0086]

[0087] Where, ΔW p (x, y) The residual error ΔW of the control surface shape in the actual state measured by the wavefront sensor system real (x, y) and the calculated control surface shape residual ΔW under the ideal state ideal Subtract (x,y) to get it.

[0088] Step 4: According to the matrix equation of formula (15), the offset coefficient k of each influence function can be solved: n , according to the offset coefficient k n Update each influence function once to obtain a new influence function. The update formula is as follows:

[0089] I2 n (x, y) = k n I n (x,y) (16)

[0090] Among them, I2 n (x, y) is the new influence function obtained after updating.

[0091] At this point, one round of regulation is completed, and the influence function is updated. Then return to step 2, and use the new influence function as the initial standard influence function in the next round of regulation. In each subsequent round of regulation, repeat steps 2 to 4 to iteratively update the influence function after each regulation to ensure the accuracy of the influence function and the accuracy of the adjustment amount calculation, and finally achieve the closed-loop solution of the correction amount of the wavefront corrector.

[0092] The present invention improves the problems existing in the existing adjustment amount calculation algorithm and has the following advantages over Algorithm 1:

[0093] 1. In order to solve the problem that the influence function of the actuator unit may be offset during use, the present invention adopts an iteratively updated influence function. After each control process, the influence function is iteratively updated in real time according to the control residual to ensure the accuracy of the influence function, thereby improving the accuracy of the calculation of the active optical correction amount;

[0094] 2. In view of the problem that the measurement and calibration methods of traditional influence functions are complicated and time-consuming, the present invention reversely calculates the influence function offset by the difference between the actual control residual and the ideal control residual, and achieves the effect of calculating all influence function offsets once. This avoids the time consumption caused by the need to recalibrate the influence function during use;

[0095] 3. The present invention adopts a method for obtaining the gradient of the influence function, screens out points where the gradient produces mutations, and fills in the abnormal points with the closest non-outlier values, thereby reducing the error introduced in the actual measurement of the influence function.

[0096] Compared with Algorithm 2, it has the following advantages:

[0097] 1. For the problem that the solution result is not the optimal solution, the present invention adopts a secondary solution strategy to achieve the optimal solution based on the relationship curve between the number of solutions and the solution accuracy;

[0098] 2. The present invention takes into account the law of environmental disturbances received by the space optical system, and has a good correction effect on the deviation of the influence function caused by regular temporal low-frequency disturbances, such as orbital periodicity, radiation periodicity and other disturbances;

[0099] 3. Compared with deep learning methods, the model does not require a large amount of data set training.

[0100] The mirror of a space optical camera is subject to a variety of disturbances in the environment. The characteristics of the disturbances can be divided into two categories: regular disturbances and random disturbances. Regular disturbances mainly include:

[0101] Thermal deformation: Temperature changes in the space environment can cause thermal expansion and contraction of optical components, thus affecting the shape of the mirror. This disturbance usually has a certain regularity, such as the temperature difference between day and night or orbital periodicity;

[0102] Radiation pressure: Solar radiation and other electromagnetic radiation exert pressure on the mirror, causing the shape to change. The intensity and direction of these radiations also have regularity.

[0103] Mechanical vibration: The satellite's attitude control, launch process or other mechanical movements will cause vibrations in the mirror, which usually follow a specific frequency and amplitude.

[0104] Regular disturbances can be modeled as periodic aberrations of different frequencies.

[0105] Random disturbances are generally manifested as noise inside the system and the influence of random airflow or stray radiation in the environment. They can be modeled as random noise disturbances.

[0106] With respect to the displacement error of the actuator unit following the voltage cycle, the present invention carries out the following simulation analysis of the correction capability:

[0107] The response curve of the actuator unit displacement to voltage is as follows: Figure 3 As shown, it can be seen that the piezoelectric ceramics of the actuator unit have nonlinear response characteristics, resulting in a certain drift in the standard position during use as the voltage increases and decreases. Since the form of aberration is determined by the form of disturbance in the system, and the form of disturbance in the system has a periodic law, the form of aberration also has a periodic change law. When adjusting these specific forms of combined aberrations, due to the nonlinear response characteristics of the actuator unit, the return error generated also changes according to the law.

[0108] Therefore, the offset of the influence function is positively correlated with the offset of the error in the environment. The offset error of the influence function is also defined to be periodic, and its period is consistent with the aberration period. Define a periodic error similar to a trigonometric function, perform multiple rounds of aberration correction, and calculate the correction residual RMSE under ideal conditions (no drift of the influence function), actual conditions (drift of the influence function), and after adjustment (using iteratively updated influence functions). The results are Figure 4 As shown in Figure 1, it can be seen that the surface shape residual calculated using the iteratively updated influence function is very close to the ideal value, which is a significant improvement compared to the existing methods.

[0109] The calculation algorithm of the present invention is used to correct the deviation of a certain influence function generated by simulation in real time, and the correction result is ideal. The error model that can be actually applied by the present invention is not limited to the above-mentioned example model.

[0110] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0111] The above-mentioned embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A closed-loop solution method for active optical correction based on iterative update of influence functions, characterized in that: The following steps are involved: Step 1: Collect the initial influence function I0 of each actuator unit in the current wavefront corrector n (x, y), and the initial influence function I0 n (x, y) is used to replace the abnormal point and obtain the initial standard influence function I of each actuator unit n (x, y), where n is the number of the actuator unit, n = 0, 1, 2, ..., N; Step 2: Use the wavefront sensing system to detect the system wavefront aberration W1(x, y) before correction, and according to the system wavefront aberration W1(x, y) in the ideal state and the initial standard influence function I of each actuator unit n (x, y) and the voltage v applied to each actuator unit n The matrix equation between is solved to obtain the voltage v1 that should be applied to each actuator unit. n , and then obtain the control surface shape and control surface shape residual ΔW in the ideal state respectively ideal (x,y), and according to the voltage n1 obtained by the solution n It is applied to the corresponding actuating unit to control the mirror surface to undergo corresponding deformation; Step 3: The wavefront sensing system detects the system wavefront aberration again to obtain the control surface shape and the control surface shape residual ΔW in the actual state. real (x, y), and the control surface shape residual ΔW in the actual state real The residual error ΔW between (x, y) and the ideal control surface shape ideal (x, y) is subtracted to obtain ΔW p (x,y), the formula is as follows: ΔW p (x,y)=ΔW real (x,y)-ΔW ideal (x,y) (13) Substitute the calculation expression of each part of the control surface residual into it and get: Among them, k n is the offset coefficient of the nth influence function; Step 4: According to ΔW shown in formula (14) p The relationship between (x, y) and the influence function used in this round of regulation is solved to calculate the offset coefficient of each influence function, and each influence function is updated once according to the offset coefficient to obtain a new influence function, and then return to step 2, and use the new influence function as the initial standard influence function in the next round of regulation, and repeat steps 2 to 4 to realize the closed-loop solution of the correction amount of the wavefront corrector.

2. The closed-loop solution method for active optical correction based on iterative update of influence function according to claim 1 is characterized in that: In step 1, the process of replacing outliers in the influence function includes the following steps: Step 1.1: Calculate each initial influence function I0 n The image gradient information of (x, y) is calculated as follows: (Gx,Gy)=Gradient(I0 n (x,y)) (3) Among them, Gx and Gy are the gradients in the x-direction and y-direction respectively; Step 1.2: Identify the mutation point in the gradient map as an outlier, record the coordinates of the outlier (x0, y0), and use the nearest point interpolation to replace the value at the outlier. The formula is as follows: I0 n (x0,y0)=Nearest(x0,y0) (5) Among them, T0 is the outlier threshold, and Nearest(x0,y0) is the nearest point interpolation function of the outlier. Step 1.3: After all abnormal points are replaced, the initial standard influence function I of the actuator unit is obtained. n (x, y).

3. The closed-loop solution method for active optical correction based on iterative update of influence function according to claim 1 or 2, characterized in that: The initial influence function of each actuator unit is obtained by interferometer measurement.

4. The closed-loop solution method for active optical correction based on iterative update of influence function according to claim 1 or 2, characterized in that: The voltage v1 that should be applied to each actuator unit is obtained by using the least squares method twice. n .

5. The closed-loop solution method for active optical correction based on iterative update of influence function according to claim 4 is characterized in that: Use genetic algorithm or simulated annealing algorithm to replace the least squares method.

6. The closed-loop solution method for active optical correction based on iterative update of influence function according to claim 1 or 2, characterized in that: The formula for updating each influence function according to the offset coefficient is as follows: I2 n (x,y)=k n I n (x,y) (16) Among them, I2 n (x, y) is the new influence function obtained after updating.

Images