Adaptive wavefront control method and system based on parameter identification and active disturbance rejection

Abstract

The application relates to an adaptive wavefront control method and system based on parameter identification and active disturbance rejection. The method comprises a digital architecture adopting PC data processing and FPGA control. The PC data processing module is used to perform complex calculations such as variable forgetting factor recursive least squares, is responsible for dimensionality reduction decoupling and high-precision estimation of disordered data, and utilizes the high-speed parallel processing capability of the FPGA control system to directly execute an active disturbance rejection control algorithm at the bottom layer to offset external interference. The hardware-software collaborative design effectively combines the accuracy of the complex algorithm and the real-time performance of the bottom-layer hardware, thereby significantly enhancing the anti-interference capability and response bandwidth of the system.

Description

Technical Field

[0001] This application relates to the fields of microelectromechanical systems and adaptive optics, and in particular to an adaptive wavefront control method and system based on parameter identification and active disturbance rejection. Background Technology

[0002] Microelectromechanical systems (MEMS) deformable mirrors, as miniature actuators capable of precise wavefront phase compensation, offer advantages such as low power consumption, fast response speed, high unit integration, and no hysteresis effect. Therefore, they are widely used in fields such as astronomical imaging, biomedical imaging, laser beam shaping, and space optical communication. Specific applications include correcting aberrations caused by atmospheric turbulence, improving microscope imaging resolution, and ensuring the stability of optical communication links.

[0003] Existing wavefront control technologies have significant shortcomings in practical engineering applications. In real-world working environments, the beam is affected by complex dynamic wavefront distortions and ambient temperature drift. Existing control systems often struggle to process this complex wavefront slope data, resulting in delayed control command issuance and an inability to achieve real-time, accurate correction. Furthermore, traditional analog control circuits are not only inflexible but also highly susceptible to electronic noise and ambient temperature drift. When faced with complex wavefront reconstruction and adaptive control algorithms, traditional architectures often fall short, causing the system to easily lose closed-loop stability under harsh operating conditions. Summary of the Invention

[0004] Therefore, it is necessary to provide an adaptive wavefront control method and system based on parameter identification and self-disturbance rejection that can improve the closed-loop stability of the system, in order to address the above-mentioned technical problems.

[0005] An adaptive wavefront control method based on parameter identification and active disturbance rejection, the adaptive wavefront control method based on parameter identification and active disturbance rejection includes: Step S1: Obtain the distorted wavefront slope vector after reflection from the MEMS deformable mirror through the optical path detection module, and send the distorted wavefront slope vector to the PC data processing module for Zernike mode decomposition to obtain the mode coefficients; Step S2: Based on the pattern coefficients, the PC data processing module recursively estimates the optimal pattern coefficients using recursive least squares with a variable forgetting factor, and outputs the optimal pattern coefficients. Step S3: The PC data processing module performs response matrix fitting analysis on the target voltage value based on the optimal mode coefficient and temperature compensation coefficient, and sends the target voltage command to the FPGA control module based on the target voltage value; Step S4: The FPGA control module combines the target voltage value in the target voltage command and the working status data collected by the status feedback module, runs the active disturbance rejection control algorithm to determine the driving voltage value, generates a voltage control command based on the driving voltage value, and sends the voltage control command to the high voltage multi-channel driving module; Step S5: The high-voltage multi-channel drive module generates a corresponding drive voltage to drive the MEMS deformable mirror to deform according to the received voltage control command.

[0006] An adaptive wavefront control system based on parameter identification and self-disturbance rejection includes: a MEMS deformable mirror, an optical path detection module, a PC data processing module, an FPGA control module, a state feedback module, and a high-voltage multi-channel drive module; The MEMS deformable mirror is connected to the optical path detection module, the high-voltage multi-channel drive module, and the state feedback module. The optical path detection module is connected to the PC data processing module. The PC data processing module is connected to the FPGA control module. The FPGA control module is connected to the high-voltage multi-channel drive module and the state feedback module. The high-voltage multi-channel drive module is connected to the state feedback module. The MEMS deformable mirror is used to reflect the distorted wavefront to the optical path detection module; The optical path detection module is used to acquire the distorted wavefront slope vector after reflection from the MEMS deformable mirror, and send the distorted wavefront slope vector to the PC data processing module. The PC data processing module is used to perform Zernike mode decomposition on the distorted wavefront slope vector to obtain mode coefficients. Based on the mode coefficients, the optimal mode coefficients are recursively estimated using recursive least squares with variable forgetting factor, and the optimal mode coefficients are output. The target voltage value is then analyzed by fitting the response matrix based on the optimal mode coefficients and the temperature compensation coefficients. Finally, the target voltage command is sent to the FPGA control module based on the target voltage value. The FPGA control module is used to combine the target voltage value in the target voltage command and the working status data collected by the status feedback module, run the active disturbance rejection control algorithm to determine the driving voltage value, generate a voltage control command based on the driving voltage value, and send the voltage control command to the high voltage multi-channel driving module. The high-voltage multi-channel drive module is used to generate a corresponding drive voltage to drive the deformation of the MEMS deformable mirror according to the received voltage control command. The status feedback module is used to monitor the temperature of the MEMS deformable mirror in real time and send the temperature of the MEMS deformable mirror to the FPGA control module for calculation of the temperature compensation coefficient.

[0007] The beneficial effects of this application are: (1) By introducing a variable forgetting factor recursive least squares, the system can dynamically adjust the forgetting factor according to the residual size, converge quickly when the wavefront distortion changes drastically, and maintain high-precision estimation in steady state, thus optimizing the problem that traditional RLS cannot balance speed and accuracy in time-varying environments. (2) The system introduces temperature compensation. The temperature of the MEMS deformable mirror is monitored in real time through the state feedback module, and the response matrix and the control gain of the self-disturbance rejection are dynamically corrected. This effectively solves the problems of material stiffness drift and actuator response sensitivity change caused by temperature changes, and ensures the stability of the system under different thermal environments. (3) The PC data processing module is responsible for complex Zernike mode decomposition and RLS-VF parameter identification, while the FPGA control module is responsible for the parallel computation of the underlying active disturbance rejection control algorithm. This architecture leverages the flexibility of the PC data processing module in processing complex algorithms and the advantages of the FPGA control module in high-speed parallel processing and low-latency response, thus achieving an effective combination of complex algorithms and real-time control. (4) The system can still maintain stable closed-loop control in dynamic environments, with strong robustness, and is suitable for scenarios with extremely high requirements for beam quality, such as high-resolution astronomical observation and free-space optical communication. Attached Figure Description

[0008] Figure 1 This is a flowchart illustrating an adaptive wavefront control method based on parameter identification and active disturbance rejection in one embodiment. Figure 2 This is a schematic diagram of the RLS-VF processing flow in one embodiment; Figure 3 This is a schematic diagram of the matrix fitting process in one embodiment; Figure 4 This is a schematic diagram of the FPGA control module in one embodiment; Figure 5 This is a schematic diagram of the structure of an adaptive wavefront control system based on parameter identification and active disturbance rejection in one embodiment; Figure 6 This is a schematic diagram of the state feedback module in one embodiment; Figure 7 This is a schematic diagram of the PC data processing module in one embodiment. Detailed Implementation

[0009] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0010] In one embodiment, such as Figure 1 As shown, an adaptive wavefront control method based on parameter identification and active disturbance rejection is provided, including the following steps: Step S1: Obtain the distorted wavefront slope vector after reflection from the MEMS deformable mirror through the optical path detection module, and send the distorted wavefront slope vector to the PC data processing module for Zernike mode decomposition to obtain the mode coefficients.

[0011] Step S2: The PC data processing module uses recursive least squares with variable forgetting factor to recursively estimate the optimal pattern coefficients based on the pattern coefficients and outputs the optimal pattern coefficients.

[0012] In Recursive Least Squares with Variable Forgetting Factor (RLS-VF), the pattern coefficients are... As the state input, dynamic optimization estimation is performed.

[0013] Among them, the mode coefficients based on Zernike mode decomposition Recursive least squares applied to a variable forgetting factor enables high-precision recursive estimation of time-varying wavefront perturbations. An initial error covariance matrix can be set. for and initial mode coefficients After setting it as the zero vector, the iteration is performed according to the following steps: First, calculate the prior residual. Reflecting the accuracy of the current model's forecasts, The slope vector For the mapping matrix, This is the estimated value of the pattern coefficients from the previous time step. Secondly, the forgetting factor is dynamically updated based on the prior residuals. This is used to determine the output of historical information that has been forgotten. This is the sensitivity adjustment coefficient, and its adjustment range can be 0.01~10; This is the preset lower limit of the forgetting factor. This is the preset upper limit of the forgetting factor. and They are usually taken as 0.95 and 1 respectively. This involves exponential operations on the function. Finally, the gain vector is updated sequentially. Optimal mode coefficient estimates And the covariance matrix used for the next iteration. .

[0014] Step S3: The PC data processing module performs response matrix fitting analysis on the target voltage value based on the optimal mode coefficient and temperature compensation coefficient, and sends the target voltage command to the FPGA control module based on the target voltage value.

[0015] Step S4: The FPGA control module combines the target voltage value in the target voltage command with the working status data collected by the status feedback module, runs the active disturbance rejection control algorithm to determine the driving voltage value, generates a voltage control command based on the driving voltage value, and sends the voltage control command to the high-voltage multi-channel drive module.

[0016] The operating status data may include the temperature of the MEMS deformable mirror and the high-voltage amplifier circuit of the high-voltage multi-channel drive module, the current of the high-voltage multi-channel drive module, the feedback voltage of the high-voltage amplifier circuit of the high-voltage multi-channel drive module, and power consumption. In step S4, the FPGA control module combines the target voltage value in the target voltage command with the temperature of the MEMS deformable mirror and the feedback voltage of the high-voltage amplifier circuit of the high-voltage multi-channel drive module collected by the status feedback module to run an active disturbance rejection control algorithm to determine the drive voltage value.

[0017] The FPGA control module receives temperature data from the status feedback module, calculates the temperature compensation coefficient, and applies it to the control gain in the active disturbance rejection control algorithm.

[0018] The final target voltage value, obtained by fitting the response matrix of the FPGA control module and the PC data processing module, along with the temperature compensation coefficient, is used by the active disturbance rejection control algorithm to calculate the final drive voltage value output to the MEMS deformable mirror actuator.

[0019] Step S5: The high-voltage multi-channel drive module generates a corresponding drive voltage to drive the MEMS deformable mirror to deform according to the received voltage control command.

[0020] In one embodiment, the steps of Zernike pattern decomposition include: Step S11: Obtain the distorted wavefront slope vector ,in, This is the index for the current sampling time, where K is the number of sub-apertures of the wavefront sensor, and the superscript T indicates transpose. For the wavefront function within the sub-aperture of the corresponding wavefront sensor The partial derivatives satisfy , ; The slope of the x-axis represents the local region of the wavefront. The slope of the y-axis represents the local region of the wavefront; Step 12: Establish the slope vector by expanding the wavefront function into a linear combination of Zernike polynomials. in; For the mapping matrix, For model coefficients, To measure the noise vector, These represent the specific mode coefficient matrices corresponding to Zernike polynomials of orders 1 to M, respectively. The expression for the wavefront function expanded into a linear combination of Zernike polynomials is: ;in, For Zernike polynomial basis functions, each order Represents a specific optical aberration mode. For at any time of Zernike mode coefficients; The total order of the decomposition is given by .

[0021] M is usually taken as the first 35 or the first 65 orders.

[0022] In one embodiment, such as Figure 2 As shown, step S2 specifically includes: Step S21: Convert the mode coefficients As the state input, establish the criterion function. ,in, for The residual slope vector measured at time step; It is a forgetting factor, and ; Step S22: Calculate the time based on the criterion function. a priori residuals ,in, These are the estimated model coefficients from the previous time step; Step S23: According to time Prior residual update time Forgetting factor ,in, This is the sensitivity adjustment coefficient. This is the preset lower limit of the forgetting factor. This is the preset upper limit of the forgetting factor. For exponential function operations, its independent variable is ; Step S24: Update the gain vector based on the forgetting factor at time k. ,in, For a moment The error covariance matrix; Step S25: Update the covariance matrix based on the gain vector and forgetting factor. ; For a moment The error covariance matrix, which serves as the benchmark for the next iteration; Step S26: Update the mode coefficient estimates based on the gain vector and prior residuals. ,in, These are the estimated model coefficients from the previous time step. Let be the gain vector at time k; Let k be the prior residual at time k; For the updated mode coefficient estimates at time k, As the optimal mode coefficient at time k.

[0023] After completing the system initialization of the mode coefficients and error covariance matrix, the criterion function is first established. ,in for The residual slope vector measured at time t. Forgetting factor ( The adaptive logic derivation of the variable forgetting factor is then performed, calculating the prior residuals. Reflecting the accuracy of the current model's forecasts, These are the estimated model coefficients from the previous time step.

[0024] Secondly, update the forgetting factor. This is used to determine the output of historical information that has been forgotten. and Typically, these are set to 0.95 and 1.0. Then, the gain vector is updated. Subsequent During the update, the system assigns higher weights to newly collected data. Then, the covariance matrix is ​​updated. This covariance matrix serves as the baseline for the next iteration. Finally, the new mode coefficient estimates are obtained. calculate, The "best guess" of the wavefront distortion state, i.e. the optimal mode coefficients output by RLS-VF, is used to prepare for the fitting of the response matrix.

[0025] In one embodiment, establishing the mapping relationship between each actuator of the MEMS deformable mirror and the output wavefront mode through response matrix fitting to obtain the target voltage value requires the following target voltage value analysis steps: Step S31: Establish the slope generated by the mirror deformation of the MEMS deformable mirror. Relationship with actuator voltage ,in, The actuator influence matrix; This is the voltage vector applied to each actuator; Step S32: Adjust the slope of the mirror deformation of the MEMS deformable mirror. Exactly offset by the optimal mode coefficients The distortion slope described , making = The voltage vector applied to each actuator at time k is solved using the least squares method. And from this, the baseline response matrix is ​​derived. , used for pre-calculation and storage; Step S33: Introduce a temperature compensation coefficient Calculate the actual response matrix at the current temperature. ;in, This is the coefficient of mirror sensitivity as a function of temperature, and its value is determined by the material of the currently manufactured MEMS deformable mirror. The temperature of the MEMS deformable mirror is collected in real time by a temperature sensor. This is the initial temperature of the mirror surface; Step S34: Introduce regularization term Calculate voltage increment The target voltage value is obtained as follows: ,in, It is the identity matrix. This is the voltage vector applied to each actuator at the previous moment.

[0026] Among them, such as Figure 3 As shown, after RLS-VF outputs the optimal mode coefficients, it first performs input and initialization to determine the slope generated by the mirror deformation of the MEMS deformable mirror. Exactly offset by the optimal mode coefficients The distortion slope described This ensures that the mirror distortion and the aberration are perfectly matched. That is, the formula... and The formulas are equal, where the matrix is ​​equal. The actuator influence matrix is ​​then calculated, followed by static reference calculations, and the voltage vector is calculated using the least squares method. And the baseline response matrix is ​​obtained. This is used for pre-calculation and storage, avoiding large-scale matrix multiplication in real time. Then, thermal compensation correction is introduced to consider the influence of ambient temperature on the physical properties of MEMS, and a formula is used to adjust the reference response. Response matrix The actual response matrix at the current temperature is corrected. .

[0027] Secondly, stable incremental control is used to maintain computational stability in the presence of noise and actuator coupling, preventing the actual response matrix from being affected. If an ill-conditioned solution is found, a smooth output solution is achieved by introducing a regularization term. Calculate the voltage increment used to compensate for the current wavefront distortion. Finally, the final voltage (i.e. the target voltage value) is output.

[0028] In one embodiment, such as Figure 4 As shown, step S4 includes: Step S41: Use the tracking differentiator of the FPGA control module to smooth the target voltage value, thus smoothing the target voltage value. The expression for the smoothed tracking signal is as follows: ; in, This is the smoothed voltage target position command. for The rate of change For the rate of change of the voltage target, for The rate of change This is the fastest control synthesis function. In the fastest control synthesis function... For the error term, in the fastest control synthesis function For the speed term, For velocity factor, For the filter factor, and The value is determined by system parameters; Step S42: Use the extended state observer of the FPGA control module to estimate the total system disturbance in real time, the expression of which is: ; in, The status feedback module monitors the feedback voltage of the output channel of the high-voltage multi-drive channel module. To Real-time tracking; For estimating the trend of voltage changes; The total disturbance estimate for the observation of the extended state is used to absorb and offset material stiffness drift, hysteresis and creep caused by temperature changes in real time. for The rate of change for The rate of change for The rate of change To output the estimated error gain, For the state estimation error gain, To estimate the error gain for the disturbance, based on the system bandwidth... Set as ; For temperature-compensated control gain, , This is the temperature compensation coefficient; It is a nonlinear function. For error variables, The width of the linear interval. For the system at the initial reference temperature The nominal control gain constant is below; This is a term used to control the influence of inputs on the rate of change of state; The nonlinear exponent for the state estimation channel. The nonlinear exponent for perturbation estimation of the channel. This is the voltage after total disturbance reduction and temperature compensation; Step S43: Generate the initial control decision voltage using the nonlinear state error feedback of the FPGA control module. Its expression is: ; ; in, For positional error, For speed error, For position error feedback gain; For speed error feedback gain; The nonlinear exponent of the position error determines how much the current surface shape of the MEMS deformable mirror differs from the target surface shape; The nonlinear exponent of the velocity error determines the sensitivity of the MEMS deformable mirror during rapid deformation. Step S44: Analyze the final control law, its expression is: ; ; in, This is the voltage after total disturbance reduction and temperature compensation; This is the driving voltage value. It is a symbolic function.

[0029] In this process, a tracking differential (TD) is used to smooth the target voltage value. In order to avoid the target voltage command issued by the PC data processing module as a step command causing instantaneous impact on the MEMS deformable mirror, the command is converted into a smooth tracking signal. A transition process is arranged to suppress system overshoot, while the differential signal of the command is extracted for trend prediction.

[0030] Specifically, the Extended State Observer (ESO) is used to combine internal uncertainties and external disturbances into a new state variable as the total disturbance. The error is combined through a nonlinear function to generate a robust driving control increment, which is then estimated and captured in real time.

[0031] Among them, nonlinear state error feedback (NLSEF) is used to... , With the observed state Comparison is performed to generate the initial control decision voltage. .

[0032] The target voltage value and temperature compensation coefficient are combined, and the FPGA control module calculates the final drive voltage output to the MEMS deformable mirror actuator through active disturbance rejection control. First, a tracking differentiator (TD) is used to smooth the target voltage value. In order to avoid the target voltage command issued by the PC data processing module as a step command causing instantaneous impact on the MEMS deformable mirror, it is converted into a smooth tracking signal.

[0033] Then, the Extended State Observer (ESO) is used to combine internal uncertainties and external disturbances into a new state variable as the total disturbance, and to estimate and capture it in real time.

[0034] Subsequently, nonlinear state error feedback (NLSEF) and linearized control input are used to compare the processed command with the observed state to generate the initial control decision voltage. .

[0035] Finally, the final control law is derived, and its expression is: , The drive voltage value is sent to the high-voltage multi-channel drive module.

[0036] The temperature compensation module (TC) receives temperature data from the status feedback module and calculates the temperature compensation coefficient.

[0037] The aforementioned adaptive wavefront control method based on parameter identification and active disturbance rejection (ADDR) performs Zernike mode decomposition on the distorted wavefront slope vector using a PC data processing module. It then employs recursive least squares with a variable forgetting factor for recursive estimation, outputting the optimal mode coefficients. Temperature compensation is then used to fit the response matrix, resulting in the target voltage value. The FPGA control module runs the ADDR algorithm, smoothing the data using a tracking differentiator, estimating the total disturbance using an extended state observer, generating control inputs through nonlinear state error feedback, and correcting the control gain using temperature compensation. Finally, it outputs a voltage control command based on the driving voltage value to control the high-voltage multi-channel drive module to generate the corresponding driving voltage to the MEMS deformable mirror. This method achieves high-precision estimation and compensation for time-varying wavefront disturbances, effectively suppressing the impact of temperature changes on the system and improving the adaptive capability and control accuracy of wavefront correction.

[0038] The adaptive wavefront control method based on parameter identification and active disturbance rejection in this application adopts a digital architecture of PC data processing and FPGA control. Specifically, the PC data processing module leverages its strengths in performing complex calculations such as recursive least squares with variable forgetting factors to reduce the dimensionality of cluttered data, decouple it, and perform high-precision prediction. Simultaneously, the high-speed parallel processing capabilities of the FPGA control system are utilized to directly execute the active disturbance rejection control algorithm at the underlying level to counteract external interference. This hardware-software co-design effectively combines the accuracy of complex algorithms with the real-time performance of the underlying hardware, thereby significantly enhancing the system's anti-interference capability and response bandwidth.

[0039] It should be understood that, although Figure 1 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1 At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0040] In one embodiment, such as Figure 5 As shown, an adaptive wavefront control system based on parameter identification and self-disturbance rejection is provided, including: a MEMS deformable mirror, an optical path detection module, a PC data processing module, an FPGA control module, a state feedback module, and a high-voltage multi-channel drive module; The MEMS deformable mirror is connected to the optical path detection module, the high-voltage multi-channel drive module, and the status feedback module. The optical path detection module is connected to the PC data processing module. The PC data processing module is connected to the FPGA control module. The FPGA control module is connected to the high-voltage multi-channel drive module and the status feedback module. The high-voltage multi-channel drive module is connected to the status feedback module.

[0041] MEMS deformable mirrors are used to reflect distorted wavefronts into the optical path detection module.

[0042] The MEMS deformable mirror receives a voltage signal from the high-voltage multi-channel drive module. An actuator generates electrostatic deformation, outputting a phase compensation signal opposite to the distorted wavefront.

[0043] The optical path detection module is used to acquire the distorted wavefront slope vector after reflection from the MEMS deformable mirror and send the distorted wavefront slope vector to the PC data processing module.

[0044] The optical path detection module receives the reflected beam from the MEMS deformable mirror via a conventional optical path, and the wavefront sensor obtains the distorted wavefront slope vector and sends it to the PC data processing module.

[0045] The optical path detection module can use a Shaker-Hartmann wavefront sensor.

[0046] The optical path detection module is connected to the optical path of the MEMS deformable mirror.

[0047] The PC data processing module is connected to the optical path detection module and the FPGA control module.

[0048] The PC data processing module can be used to solve the received distorted wavefront slope vector into a target voltage value and send it to the FPGA control module.

[0049] The PC data processing module executes Zernike mode decomposition, variable forgetting factor recursive least squares, and response matrix fitting algorithms through an embedded wavefront solver.

[0050] The PC data processing module may include: a Zernike mode decomposition unit for extracting mode coefficients, a variable forgetting factor recursive least squares (RLS-VF) unit for estimating optimal mode coefficients, a response matrix fitting unit for calculating target voltage values, and a sensor parameter display unit for visualizing state data.

[0051] The PC data processing module can establish bidirectional data interaction with the FPGA control module through serial communication, fieldbus or Ethernet communication protocols. The PC data processing module can be configured with a host computer monitoring interface for real-time display of sensor status data.

[0052] In one embodiment, the PC data processing module performs Zernike mode decomposition on the distorted wavefront slope vector to obtain mode coefficients. Based on the mode coefficients, it uses recursive least squares with a variable forgetting factor to recursively estimate the optimal mode coefficients, outputs the optimal mode coefficients, and performs response matrix fitting analysis on the target voltage value based on the optimal mode coefficients and temperature compensation coefficients. Based on the target voltage value, it sends a target voltage command to the FPGA control module.

[0053] The FPGA control module can be used to generate a driving voltage value by running an active disturbance rejection control algorithm based on the target voltage value in the target voltage command and the state feedback information. The FPGA control module may include: a tracking differentiator (TD) for smoothing the target voltage value, an extended state observer (ESO) for observing and estimating system disturbances, a nonlinear state error feedback (NLSEF) for calculating the final control increment, a control law analysis unit for analyzing the final control law, a temperature compensation module (TC) for calculating the compensation coefficient, and a sensor parameter merging unit for summarizing state data and transmitting it back to the PC data processing module.

[0054] In one embodiment, the FPGA control module combines the target voltage value and the operating status data collected by the status feedback module, runs an active disturbance rejection control algorithm to determine the drive voltage value, generates a voltage control command based on the drive voltage value, and sends the voltage control command to the high-voltage multi-channel drive module.

[0055] The FPGA control module writes the drive voltage value calculated by the active disturbance rejection control algorithm into the on-chip ARM processor for buffer storage, and then synchronously distributes it to the high-voltage multi-channel drive module.

[0056] In one embodiment, the high-voltage multi-channel drive module is used to generate a corresponding drive voltage to drive the deformation of the MEMS deformable mirror according to the received voltage control command.

[0057] The high-voltage multi-channel drive module includes a digital-to-analog converter interface and a high-voltage amplifier circuit, which is used to convert the drive voltage value into a parallel high-voltage drive signal.

[0058] The high-voltage multi-channel drive module receives the drive voltage value from the FPGA control module through the digital-to-analog converter interface and synchronously distributes it to multiple preset parallel output channels. Then, the low-voltage control signal is linearly amplified by the high-voltage amplifier circuit to generate a high-voltage drive voltage sufficient to drive the deformation of the MEMS deformable mirror unit. The physical data of the high-voltage multi-channel drive module is monitored by the status feedback module.

[0059] The status feedback module can be used to collect working status data and feed it back to the FPGA control module to form a closed loop.

[0060] The current monitor monitors the input current data of the high-voltage multi-channel drive module, the temperature sensor monitors the temperature data of the MEMS deformable mirror and the high-voltage multi-channel drive module, and the feedback voltage monitor monitors the feedback voltage data of the high-voltage multi-channel drive module. All data will be sent to the FPGA control module for processing.

[0061] In one embodiment, the status feedback module is used to monitor the temperature of the MEMS deformable mirror in real time and send the temperature of the MEMS deformable mirror to the FPGA control module for calculation of the temperature compensation coefficient.

[0062] Among them, such as Figure 6 As shown, the status feedback module may include a current monitor, a temperature sensor, and a feedback voltage monitor, which are respectively connected to the FPGA control module.

[0063] In one embodiment, such as Figure 7 As shown, the PC data processing module includes a Zernike mode decomposition unit, an RLS-VF unit, and a response matrix fitting unit; The Zernike pattern decomposition unit is used to perform Zernike pattern decomposition. The RLS-VF unit is used to recursively estimate the optimal mode coefficients based on the mode coefficients and employs recursive least squares with a variable forgetting factor, and outputs the optimal mode coefficients. The response matrix fitting unit is used to perform response matrix fitting analysis on the target voltage value based on the optimal mode coefficients and temperature compensation coefficients, and sends the target voltage command to the FPGA control module based on the target voltage value.

[0064] In one embodiment, the PC data processing module further includes a sensor parameter display unit, which is used to receive temperature, current, voltage and power consumption data processed and merged by the FPGA control module in real time, and to display them visually on the host computer monitoring interface.

[0065] In one embodiment, such as Figure 4 As shown, the FPGA control module includes: a tracking differentiator, an extended state observer, a nonlinear state error feedback, a temperature compensation module, and a control law analysis unit; The tracking differentiator is used to smooth the target voltage value, reducing the target voltage value... The expression for the smoothed tracking signal is as follows: ; in, This is the smoothed voltage target position command. for The rate of change For the rate of change of the voltage target, for The rate of change This is the fastest control synthesis function. In the fastest control synthesis function... For the error term, in the fastest control synthesis function For the speed term, For velocity factor, The filter factor; The extended state observer is used to estimate the total disturbance of the system in real time, and its expression is: ; in, The feedback voltage monitored by the status feedback module. To Real-time tracking; For estimating the trend of voltage changes; Estimate the total perturbation observed in the extended state; for The rate of change for The rate of change for The rate of change To output the estimated error gain, For the state estimation error gain, To estimate the error gain for the disturbance, based on the system bandwidth... Set as ; For temperature-compensated control gain, , This is the temperature compensation coefficient; It is a nonlinear function. For error variables, The width of the linear interval. For the system at the initial reference temperature The nominal control gain constant is below; This is a term used to control the influence of inputs on the rate of change of state; The nonlinear exponent for the state estimation channel. The nonlinear exponent for perturbation estimation of the channel. This is the voltage after total disturbance reduction and temperature compensation; Nonlinear state error feedback is used to generate the initial control decision voltage. Its expression is: ; ; in, For positional error, For speed error, For position error feedback gain; For speed error feedback gain; This is the nonlinear exponent of the position error. This is the nonlinear exponent of the velocity error; The temperature compensation module is used to calculate the temperature compensation coefficient and the temperature-compensated control gain based on the temperature of the MEMS deformable mirror, and outputs the temperature compensation coefficient and the temperature-compensated control gain. The control law analysis unit is used to analyze the final control law, and its expression is: ; ; in, This is the voltage after total disturbance reduction and temperature compensation; This is the driving voltage value. It is a symbolic function.

[0066] In one embodiment, the FPGA control module further includes: a sensor parameter merging unit; The sensor parameter merging unit is used to calculate power consumption based on the temperature, voltage, and current data sent by the status feedback module, and then merges the temperature, voltage, current, and power consumption data into the ARM buffer storage before transmitting it to the sensor parameter display of the PC data processing module for data display.

[0067] In one embodiment, the selected MEMS deformable mirror contains 256 independent actuators, and the wavefront sensor is divided into 12×12 sub-apertures. First, when an external light source is affected by factors such as atmospheric turbulence, resulting in a distorted wavefront, the MEMS deformable mirror reflects the distorted wavefront to the wavefront sensor in the optical path detection module. Subsequently, the wavefront sensor processes the distorted wavefront data and sends the distorted wavefront slope vector to the PC data processing module. The PC data processing module receives the distorted wavefront slope vector. After internal Zernike pattern decomposition algorithm, variable forgetting factor recursive least squares (RLS-VF) algorithm, and combined with temperature compensation factor from FPGA control module The target voltage value of the distortion wavefront slope vector to be compensated is obtained by fitting the response matrix. and the target voltage value The input is sent to the FPGA control module. The FPGA control module then processes the target voltage value from the PC data processing module. Active disturbance rejection control is implemented by smoothing the signal using a tracking differentiator, monitoring the total disturbance using an extended state observer, and combining nonlinear state error feedback with the tracking differentiator, extended state observer, and temperature compensation factor. Generate initial control decision voltage Temperature compensation factor The temperature data from the MEMS deformable mirror temperature sensor, received from the state feedback module, is calculated by the TC module of the FPGA control module. This is determined by the initial control decision voltage. The final drive voltage value is obtained by performing control law calculations. Drive voltage value The signal is sent to the high-voltage multi-channel drive module. The high-voltage multi-channel drive module receives the drive voltage value from the FPGA control module via a digital-to-analog converter interface. The low-voltage control signal is then synchronously distributed to multiple preset parallel output channels. A high-voltage amplifier circuit amplifies the low-voltage control signal to generate a high-voltage drive voltage sufficient to drive the deformation of the MEMS deformable mirror unit. The MEMS deformable mirror deforms to compensate for the current distortion wavefront. The state feedback module simultaneously monitors the temperature of the MEMS deformable mirror and the high-voltage amplifier circuit of the high-voltage multi-channel drive module, monitors the current of the high-voltage multi-channel drive module, and monitors the feedback voltage of the high-voltage amplifier circuit of the high-voltage multi-channel drive module. These three types of data are sent to the FPGA control module. The FPGA control module calculates the temperature compensation factor using the MEMS deformable mirror temperature data, and the channel output feedback voltage of the high-voltage amplifier circuit of the high-voltage multi-channel drive module is sent to the extended state observer. Current data is used for power consumption calculation. Furthermore, the FPGA control module sends the temperature, voltage, current, and power consumption data to the sensor parameter display unit of the PC data processing module for real-time display of sensor parameters. This completes the closed-loop and real-time display of adaptive optical wavefront control with parameter identification and active disturbance rejection control.

[0068] The sensor parameter display unit receives temperature, current, voltage, and power consumption data processed and merged by the FPGA control module in real time, and displays them visually on the host computer monitoring interface, forming a data monitoring closed loop.

[0069] The system includes a current monitor that tracks the multiple currents across various power supply voltages of the high-voltage multi-channel drive module, with the current data transmitted to the FPGA control module for processing. A temperature sensor monitors the temperature of the high-voltage amplifier circuit and the MEMS deformable mirror within the high-voltage multi-channel drive module, with the temperature data also transmitted to the FPGA control module for processing. A feedback voltage monitor tracks the feedback voltage of the output channels of the high-voltage multi-channel drive module, with the feedback voltage transmitted to the FPGA control module for further processing.

[0070] This application presents an adaptive wavefront control system based on parameter identification and active disturbance rejection, employing a digital architecture of PC data processing and FPGA control. The PC data processing module leverages its strengths in performing complex calculations such as recursive least squares with variable forgetting factors, responsible for dimensionality reduction, decoupling, and high-precision prediction of cluttered data. Simultaneously, the high-speed parallel processing capabilities of the FPGA control system are utilized to directly execute the active disturbance rejection control algorithm at the underlying level to counteract external interference. This hardware-software co-design effectively combines the accuracy of complex algorithms with the real-time performance of the underlying hardware, thereby significantly enhancing the system's anti-interference capability and response bandwidth.

[0071] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above 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.

[0072] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. An adaptive wavefront control method based on parameter identification and active disturbance rejection, characterized in that, The adaptive wavefront control method based on parameter identification and active disturbance rejection includes: Step S1: Obtain the distorted wavefront slope vector after reflection from the MEMS deformable mirror through the optical path detection module, and send the distorted wavefront slope vector to the PC data processing module for Zernike mode decomposition to obtain the mode coefficients; Step S2: Based on the pattern coefficients, the PC data processing module recursively estimates the optimal pattern coefficients using recursive least squares with a variable forgetting factor, and outputs the optimal pattern coefficients. Step S3: The PC data processing module performs response matrix fitting analysis on the target voltage value based on the optimal mode coefficient and temperature compensation coefficient, and sends the target voltage command to the FPGA control module based on the target voltage value; Step S4: The FPGA control module combines the target voltage value in the target voltage command and the working status data collected by the status feedback module, runs the active disturbance rejection control algorithm to determine the driving voltage value, generates a voltage control command based on the driving voltage value, and sends the voltage control command to the high voltage multi-channel driving module; Step S5: The high-voltage multi-channel drive module generates a corresponding drive voltage to drive the MEMS deformable mirror to deform according to the received voltage control command.

2. The adaptive wavefront control method based on parameter identification and active disturbance rejection according to claim 1, characterized in that, The steps of Zernike pattern decomposition include: Step S11: Obtain the distorted wavefront slope vector ,in, This is the index for the current sampling time, where K is the number of sub-apertures of the wavefront sensor, and the superscript T indicates transpose. For the wavefront function within the sub-aperture of the corresponding wavefront sensor The partial derivatives satisfy , ; The slope of the x-axis represents the local region of the wavefront of the Kth sub-aperture of the wavefront sensor. The slope of the y-axis represents the local region of the wavefront of the Kth sub-aperture of the wavefront sensor. Step 12: Establish the slope vector by expanding the wavefront function into a linear combination of Zernike polynomials. in; For the mapping matrix, For model coefficients, To measure the noise vector, These represent the specific mode coefficient matrices corresponding to Zernike polynomials of orders 1 to M, respectively. The expression for the wavefront function expanded into a linear combination of Zernike polynomials is as follows: ;in, For Zernike polynomial basis functions, each order Represents a specific optical aberration mode. For at any time of Zernike mode coefficients; The total order of the decomposition is given by .

3. The adaptive wavefront control method based on parameter identification and active disturbance rejection according to claim 2, characterized in that, Step S2 includes: Step S21: Convert the mode coefficients As the state input, establish the criterion function. ,in, for The residual slope vector measured at time step; It is a forgetting factor, and ; Step S22: Calculate the time based on the criterion function. a priori residuals ,in, These are the estimated model coefficients from the previous time step; Step S23: According to the time Prior residual update time Forgetting factor ,in, This is the sensitivity adjustment coefficient. This is the preset lower limit of the forgetting factor. The upper limit of the preset forgetting factor, For exponential function operations, its independent variable is ; Step S24: Update the gain vector based on the forgetting factor at time k. ,in, For a moment The error covariance matrix; Step S25: Update the covariance matrix based on the gain vector and the forgetting factor. ; For a moment The error covariance matrix; Step S26: Update the mode coefficient estimates based on the gain vector and the prior residuals: ,in, These are the estimated model coefficients from the previous time step. For a moment The gain vector; For a moment The prior residuals; For a moment The updated model coefficient estimates will As the optimal mode coefficient at time k.

4. The adaptive wavefront control method based on parameter identification and active disturbance rejection according to claim 3, characterized in that, The steps for analyzing the target voltage value include: Step S31: Establish the slope generated by the mirror deformation of the MEMS deformable mirror. Relationship with actuator voltage ,in, The actuator influence matrix; This is the voltage vector applied to each actuator; Step S32: Adjust the slope of the mirror deformation of the MEMS deformable mirror. Offset by optimal mode coefficients The distortion slope described , making = The voltage vector applied to each actuator at time k is solved using the least squares method. The baseline response matrix is ​​obtained. ; Step S33: Introduce a temperature compensation coefficient Calculate the actual response matrix at the current temperature. ;in, This is the coefficient representing the change in mirror sensitivity with temperature. The temperature of the MEMS deformable mirror is collected in real time by a temperature sensor. This is the initial temperature of the mirror surface; Step S34: Introduce regularization term Calculate voltage increment The target voltage value is obtained as follows: ,in It is the identity matrix. This is the voltage vector applied to each actuator at the previous moment.

5. The adaptive wavefront control method based on parameter identification and active disturbance rejection according to claim 1, characterized in that, Step S4 includes: Step S41: Use the tracking differentiator of the FPGA control module to smooth the target voltage value, thus smoothing the target voltage value. The expression for the smoothed tracking signal is as follows: ； in, This is the smoothed voltage target position command. for The rate of change For the rate of change of the voltage target, for The rate of change This is the fastest control synthesis function. In the fastest control synthesis function... For the error term, in the fastest control synthesis function For the speed term, For velocity factor, The filter factor; Step S42: Use the extended state observer of the FPGA control module to estimate the total system disturbance in real time, the expression of which is: ； in, The status feedback module monitors the feedback voltage of the output channel of the high-voltage multi-drive channel module. To Real-time tracking; For estimating the trend of voltage changes; Estimate the total perturbation observed in the extended state; for The rate of change for The rate of change for The rate of change To output the estimated error gain, For the state estimation error gain, To estimate the error gain for the disturbance, based on the system bandwidth... Set as ; For temperature-compensated control gain, , This is the temperature compensation coefficient; It is a nonlinear function. For error variables, The width of the linear interval. For the system at the initial reference temperature The nominal control gain constant is below; This is a term used to control the influence of inputs on the rate of change of state; The nonlinear exponent for state estimation channel, The nonlinear exponent for perturbation estimation of the channel. This is the voltage after total disturbance reduction and temperature compensation; Step S43: Generate the initial control decision voltage using the nonlinear state error feedback of the FPGA control module. Its expression is: ； ； in, For positional error, For speed error, For position error feedback gain; For speed error feedback gain; This is the nonlinear exponent of the position error. This is the nonlinear exponent of the velocity error; Step S44: Analyze the final control law, its expression is: ； ； in, This is the voltage after total disturbance reduction and temperature compensation; This is the driving voltage value. It is a symbolic function.

6. An adaptive wavefront control system based on parameter identification and active disturbance rejection, characterized in that, include: MEMS deformable mirror, optical path detection module, PC data processing module, FPGA control module, status feedback module and high voltage multi-channel drive module; The MEMS deformable mirror is connected to the optical path detection module, the high-voltage multi-channel drive module, and the state feedback module. The optical path detection module is connected to the PC data processing module. The PC data processing module is connected to the FPGA control module. The FPGA control module is connected to the high-voltage multi-channel drive module and the state feedback module. The high-voltage multi-channel drive module is connected to the state feedback module. The MEMS deformable mirror is used to reflect the distorted wavefront to the optical path detection module; The optical path detection module is used to acquire the distorted wavefront slope vector after reflection from the MEMS deformable mirror, and send the distorted wavefront slope vector to the PC data processing module. The PC data processing module is used to perform Zernike mode decomposition on the distorted wavefront slope vector to obtain mode coefficients. Based on the mode coefficients, the optimal mode coefficients are recursively estimated using recursive least squares with variable forgetting factor, and the optimal mode coefficients are output. The target voltage value is then analyzed by fitting the response matrix based on the optimal mode coefficients and the temperature compensation coefficients. Finally, the target voltage command is sent to the FPGA control module based on the target voltage value. The FPGA control module is used to combine the target voltage value in the target voltage command and the working status data collected by the status feedback module, run the active disturbance rejection control algorithm to determine the driving voltage value, generate a voltage control command based on the driving voltage value, and send the voltage control command to the high voltage multi-channel driving module. The high-voltage multi-channel drive module is used to generate a corresponding drive voltage to drive the deformation of the MEMS deformable mirror according to the received voltage control command. The status feedback module is used to monitor the temperature of the MEMS deformable mirror in real time and send the temperature of the MEMS deformable mirror to the FPGA control module for calculation of the temperature compensation coefficient.

7. The adaptive wavefront control system based on parameter identification and self-disturbance rejection according to claim 6, characterized in that, The PC data processing module includes a Zernike mode decomposition unit, an RLS-VF unit, and a response matrix fitting unit. The Zernike pattern decomposition unit is used to perform Zernike pattern decomposition. The RLS-VF unit is used to recursively estimate the optimal mode coefficients based on the mode coefficients using recursive least squares with variable forgetting factor, and output the optimal mode coefficients. The response matrix fitting unit is used to perform response matrix fitting analysis on the target voltage value based on the optimal mode coefficient and temperature compensation coefficient, and to send the target voltage command to the FPGA control module based on the target voltage value.

8. The adaptive wavefront control system based on parameter identification and self-disturbance rejection according to claim 7, characterized in that, The PC data processing module also includes a sensor parameter display unit, which is used to receive temperature, current, voltage and power consumption data processed and merged by the FPGA control module in real time and display them visually on the host computer monitoring interface.

9. The adaptive wavefront control system based on parameter identification and self-disturbance rejection according to claim 6, characterized in that, The FPGA control module includes: a tracking differentiator, an extended state observer, a nonlinear state error feedback, a temperature compensation module, and a control law analysis unit; The tracking differentiator is used to smooth the target voltage value, reducing the target voltage value... The expression for the smoothed tracking signal is as follows: ； in, This is the smoothed voltage target position command. for The rate of change For the rate of change of the voltage target, for The rate of change This is the fastest control synthesis function. In the fastest control synthesis function... For the error term, in the fastest control synthesis function For the speed term, For velocity factor, The filter factor; The extended state observer is used to estimate the total system disturbance in real time, and its expression is: ； in, The feedback voltage monitored by the status feedback module. To Real-time tracking; For estimating the trend of voltage changes; Estimate the total perturbation observed in the extended state; for The rate of change for The rate of change for The rate of change To output the estimated error gain, For the state estimation error gain, To estimate the error gain for the disturbance, based on the system bandwidth... Set as ; For temperature-compensated control gain, , This is the temperature compensation coefficient; It is a nonlinear function. For error variables, The width of the linear interval. For the system at the initial reference temperature The nominal control gain constant is below; This is a term used to control the influence of inputs on the rate of change of state; The nonlinear exponent for the state estimation channel. The nonlinear exponent for perturbation estimation of the channel. This is the voltage after total disturbance reduction and temperature compensation; The nonlinear state error feedback is used to generate the initial control decision voltage. Its expression is: ； ； in, For positional error, For speed error, For position error feedback gain; For speed error feedback gain; This is the nonlinear exponent of the position error. This is the nonlinear exponent of the velocity error; The temperature compensation module is used to output the temperature compensation coefficient and the temperature-compensated control gain based on the temperature of the MEMS deformable mirror. The control law analysis unit is used to analyze the final control law, and its expression is: ； ； in, This is the voltage after total disturbance reduction and temperature compensation; This is the driving voltage value. It is a symbolic function.

10. The adaptive wavefront control system based on parameter identification and self-disturbance rejection according to claim 9, characterized in that, The FPGA control module also includes: a sensor parameter merging unit; The sensor parameter merging unit is used to calculate power consumption based on the temperature, voltage, and current data sent by the status feedback module, and then merge the temperature, voltage, current, and power consumption data into the ARM buffer storage before transmitting it to the sensor parameter display of the PC data processing module for data display.

Images