Diaphragm aperture adaptive control method and system

Abstract

The invention provides a diaphragm aperture adaptive control method and system, and relates to the technical field of diaphragm control. According to the method, the LuGre model parameters are accurately obtained through the step-by-step identification technology, and the difficulty that friction parameters are difficult to directly measure is overcome; further introducing a Kriging model to learn and compensate the prediction error of the physical model; and finally, a composite predictor fusing a physical mechanism and data driving is constructed, and closed-loop adaptive control is realized in combination with sensor feedback. The control precision of the aperture of the diaphragm and the system reliability are remarkably improved, and the method is suitable for the field of precise optical instruments and high-end equipment.

Description

Technical Field

[0001] This invention relates to the field of aperture control technology, specifically to an adaptive aperture control method and system. Background Technology

[0002] As a core component in optical imaging and illumination systems, the aperture stop's precise size control directly determines image quality and luminous flux regulation performance. Currently, aperture stops are mostly driven by stepper motors or ultrasonic motors through reduction mechanisms and moving rings to open and close blades, thereby controlling the aperture size. However, this drivetrain system suffers from strong nonlinear friction, including static friction, dynamic friction, the Stribeck effect, pre-slip, hysteresis, and external disturbances. This makes it difficult for traditional linear model-based control methods to achieve high-precision aperture positioning, resulting in problems such as poor repeatability, slow response, and susceptibility to disturbances.

[0003] Existing aperture control schemes mainly include traditional PID control and feedforward compensation based on linear friction models. However, PID control is prone to overshoot, oscillation, and steady-state errors when dealing with nonlinear friction; linear friction models cannot accurately describe complex friction dynamics, resulting in limited compensation effectiveness. While current research has employed dynamic friction models such as LuGre for feedforward compensation, the relevant parameters of the LuGre model are difficult to measure directly, and even if they could be measured, the implementation cost is very high. Furthermore, drift may occur in practical applications, leading to model mismatch and a decrease in compensation effectiveness. Summary of the Invention

[0004] Purpose of the invention: The first purpose of this invention is to provide an aperture adaptive control method based on the fusion of LuGre and Kriging models to achieve fast and accurate response. The second purpose is to provide an aperture adaptive control system.

[0005] Technical solution: An adaptive aperture control method, comprising the following steps:

[0006] S1. Determine the static and dynamic parameter sets of the LuGre dynamic friction model through a step-by-step identification strategy. Use the rotor input pulse number of the drive motor that controls the aperture as the input of the LuGre dynamic friction model and the aperture prediction value as the output of the LuGre dynamic friction model to construct the aperture prediction basic model.

[0007] S2. Collect the rotor input pulse count of the control aperture and the corresponding aperture diameter measurement value to construct a training dataset;

[0008] S3. Input the training dataset into the aperture prediction base model to obtain the deviation between the aperture prediction value and the aperture measurement value corresponding to the same rotor input pulse number. Use the deviation value to train an error compensator based on the Kriging surrogate model.

[0009] S4. Combine the basic model for aperture prediction with the error compensator to construct a joint prediction model. The input value of the joint prediction model is the number of rotor input pulses, and the output is the final predicted value of the aperture after deviation correction.

[0010] S5. Set the target aperture diameter, calculate the required number of rotor input pulses using the joint prediction model, execute the number of rotor input pulses by the drive motor, calculate the aperture diameter measurement value and compare it with the target aperture diameter. If the error is greater than the set threshold, then correct the number of rotor input pulses and update the training dataset to achieve adaptive aperture control.

[0011] Specifically, the step-by-step identification strategy for determining the static and dynamic parameter sets of the LuGre dynamic friction model includes: identifying static parameters based on constant-speed experimental data, identifying dynamic parameters based on variable-speed experimental data and the static parameter set, refining the dynamic parameter identification by updating the bristle elastic deformation, and finally determining the static and dynamic parameter sets.

[0012] Specifically, identifying static parameters based on constant-rate experimental data includes:

[0013] Data on the aperture of the drive motor at different constant speeds are collected. Based on the Stribeck friction model, output parameters corresponding to each combination of static parameters and the speed of the drive motor are constructed. The output parameters are the difference between the actual value of steady-state friction force and the calculated value of steady-state friction force. The absolute value of the output parameters is minimized through a global optimization algorithm to obtain the corresponding static parameters. The static parameters include Coulomb friction force, maximum static friction coefficient, Stribeck characteristic velocity, and viscous friction coefficient.

[0014] Specifically, the Stribeck friction model formula is as follows:

[0015]

[0016] In the formula: For steady-state friction, Coulomb friction, The maximum static friction coefficient, For relative velocity, Stribeck characteristic velocity, is the coefficient of viscous friction.

[0017] Specifically, the identification of dynamic parameters based on variable speed experimental data and static parameter sets includes: based on the obtained static parameters, collecting aperture data of the drive motor during variable speed motion; constructing the output parameters corresponding to each set of dynamic parameter combinations at a certain moment during variable speed motion based on the dynamic LuGre friction model; the output parameter is the difference between the actual value of dynamic friction and the calculated value of dynamic friction; minimizing the absolute value of the output parameter through a global optimization algorithm to obtain the corresponding dynamic parameters, including bristle stiffness and bristle damping coefficient.

[0018] Specifically, the dynamic LuGre friction model formula is as follows:

[0019]

[0020] In the formula: For dynamic friction, For bristle stiffness, This is the bristle damping coefficient. The coefficient of viscous friction is... For the deformation of the mane, This refers to relative velocity.

[0021] Specifically, improving dynamic parameter identification by updating the elastic deformation of the mane includes: designing a nonlinear observer, as shown in the following formula:

[0022]

[0023]

[0024] In the formula: This is an estimate of the mane deformation. Based on The predicted value of friction is obtained. This is the measured value of friction force. For observer gain; This represents the steady-state deformation.

[0025] By embedding the nonlinear observer into the dynamic parameter identification process, the dynamic parameters can be improved.

[0026] Specifically, the formula for the final predicted aperture value is as follows:

[0027]

[0028] In the formula: This is the final predicted value for the aperture. The predicted aperture value of the grating is obtained from the basic model for aperture prediction. This is the deviation value obtained from the Kriging surrogate model.

[0029] Specifically, step S2 includes: connecting the rotor of the drive motor that controls the aperture diameter to a hollow shaft optical encoder, and using the hollow shaft optical encoder to collect the rotor input pulse number and the corresponding aperture diameter measurement value.

[0030] The present invention also provides an adaptive aperture control system, comprising:

[0031] Basic physical model construction module: used to determine the static parameter set and dynamic parameter set of LuGre dynamic friction model through a step-by-step identification strategy. The rotor input pulse number of the drive motor that controls the aperture is used as the input of LuGre dynamic friction model, and the aperture prediction value is used as the output of LuGre dynamic friction model to construct the basic model for aperture prediction.

[0032] Data acquisition module: used to collect the rotor input pulse count of the control aperture and the corresponding aperture diameter measurement value to build a training dataset;

[0033] Error compensator construction module: It is used to input the training dataset into the aperture prediction base model, obtain the deviation value between the aperture prediction value and the aperture measurement value corresponding to the same rotor input pulse number, and use the deviation value to train the error compensator based on the Kriging surrogate model;

[0034] Joint prediction model construction module: used to combine the basic model of aperture prediction with the error compensator to build a joint prediction model. The input value of the joint prediction model is the number of rotor input pulses, and the output is the final predicted value of aperture after deviation correction.

[0035] Aperture closed-loop control module: Used to set the target aperture diameter, calculate the required number of rotor input pulses using a joint prediction model, drive the motor to execute the number of rotor input pulses, calculate the aperture diameter measurement value and compare it with the target aperture diameter. If the error is greater than the set threshold, the number of rotor input pulses is corrected and the training dataset is updated to achieve adaptive aperture control.

[0036] Beneficial effects: Compared with the prior art, the significant effect of this invention is that it focuses on the deep integration of the aperture and the tribological model, combining the physical mechanism of the LuGre model with the spatial statistical prediction capability of the Kriging model. By bridging the two through friction compensation, a precise mapping relationship between pulse number → friction state → aperture is established, realizing a reliable connection between the physical model and the dynamic control of the aperture. By embedding a joint prediction algorithm in the main controller, the control algorithm is self-iteratively updated every hundred control cycles, giving the control method strong environmental adaptability and realizing closed-loop adaptive control, ensuring that the aperture mechanism can adjust the aperture stably and rapidly under high friction and nonlinear conditions. Attached Figure Description

[0037] Figure 1 This is a flowchart of the method in Embodiment 1 of the present invention.

[0038] Figure 2 This is a schematic diagram comparing the step response in Embodiment 3 of the present invention.

[0039] Figure 3 This is a three-dimensional surface schematic diagram showing the strong nonlinear mapping relationship between pulse number, friction torque and aperture in Embodiment 3 of the present invention. Detailed Implementation

[0040] A preferred embodiment of the present invention will be further described below with reference to the accompanying drawings.

[0041] Example 1

[0042] Please see Figure 1 As shown, this embodiment provides an adaptive aperture control method, including the following steps:

[0043] S1. The static parameter set and dynamic parameter set of the LuGre dynamic friction model are determined by a step-by-step identification strategy. The rotor input pulse number of the drive motor that controls the aperture is used as the input of the LuGre dynamic friction model, and the aperture prediction value is used as the output of the LuGre dynamic friction model to construct the basic model for aperture prediction.

[0044] In this embodiment, the step-by-step identification strategy includes: identifying static parameters based on constant-speed experimental data, identifying dynamic parameters based on variable-speed experimental data and the static parameter set, improving the identification of dynamic parameters by updating the bristle elastic deformation, and finally determining the static parameter set and the dynamic parameter set. The details are explained below.

[0045] Data on the aperture of the drive motor at different constant speeds were collected. At constant speed, the bristle deformation of the system remained unchanged. The steady-state bristle deformation was solved using the bristle deformation dynamics equation.

[0046]

[0047] In the formula: The rotation angle of the actuator, For relative velocity, For bristle stiffness, This represents the steady-state deformation. This represents the steady-state bristle deformation.

[0048] The steady-state bristle deformation and its change and Substituting the expression into the friction force input equation, we obtain the Stribeck friction model formula:

[0049]

[0050] In the formula: For steady-state friction, Coulomb friction, The maximum static friction coefficient, For relative velocity, Stribeck characteristic velocity, is the coefficient of viscous friction.

[0051] Based on this, construct each set of static parameter combinations With motor speed Corresponding output parameters:

[0052]

[0053] In the formula: For output parameters, This represents the true value of steady-state friction. This is the calculated value of steady-state friction force.

[0054] Construct a target evaluation function based on output parameters. :

[0055]

[0056] Minimize using a global optimization algorithm Accurately identify the corresponding static parameters at this time. .

[0057] Based on the known static parameters, aperture data of the drive motor during its variable speed motion are collected. The dynamic LuGre friction model is used as the basis, and the model formula is as follows:

[0058]

[0059] In the formula: For dynamic friction, For bristle stiffness, This is the bristle damping coefficient. The coefficient of viscous friction is... For the deformation of the mane, This refers to relative velocity.

[0060] Then construct each set of dynamic parameter combinations. The output parameters at a certain moment during the variable speed motion:

[0061]

[0062] In the formula: For output parameters, For a moment, This represents the true value of dynamic friction. This is the calculated value for dynamic friction force.

[0063] Construct a target evaluation function based on this output parameter. :

[0064]

[0065] Minimize using a global optimization algorithm Accurately identify the corresponding dynamic parameters at this time. .

[0066] By updating the elastic deformation of the mane to improve dynamic parameter identification, this embodiment designs a nonlinear observer, as shown in the following formula:

[0067]

[0068]

[0069] In the formula: This is an estimate of the mane deformation. Based on The predicted value of friction is obtained. This is the measured value of friction force. For observer gain; This represents the steady-state deformation.

[0070] By embedding a nonlinear observer into the dynamic parameter identification process, a target evaluation function is constructed to improve the dynamic parameters.

[0071] Substituting the static and dynamic parameter sets obtained above into the LuGre model serves as the fundamental physical model for aperture prediction. Its input is the pulse number N, and its output is the preliminary predicted aperture value D. It is assumed that the pulse number N is related to the linear displacement or angle of the actuator. The relationship is linear: ( (where D is the transmission coefficient), the force generated by the actuator needs to overcome friction and elastic forces. Solving the system dynamics equations of the LuGre model yields the preliminary predicted aperture value D.

[0072] S2. Collect the rotor input pulse count of the control aperture and the corresponding aperture diameter measurement value to construct a training dataset.

[0073] In this embodiment, as a general approach, the rotor of the drive motor controlling the aperture needs to be connected to a hollow shaft optical encoder, and the hollow shaft optical encoder is used to collect the rotor input pulse number and its corresponding actual aperture measurement value.

[0074] S3. Input the training dataset into the aperture prediction base model to obtain the deviation between the aperture prediction value and the aperture prediction value corresponding to the same rotor input pulse number. Use the deviation value to train an error compensator based on the Kriging surrogate model.

[0075] In this embodiment, the Kriging surrogate model is used as a corrector. The Lugre model with N input pulses calculates the predicted aperture value D and other measurable variables. The output is the predicted aperture value D and the predicted aperture value for each pulse number N. Deviation between :

[0076]

[0077] Using deviation value Train a Kriging surrogate model. The Kriging model is a Gaussian process regression that assumes the data follows a Gaussian process and provides a predicted value and its uncertainty. Given a new input vector, the Kriging model predicts the following value: This is the condition vector.

[0078] S4. Combine the basic model for aperture prediction with the error compensator to construct a joint prediction model. The input value of the joint prediction model is the number of rotor input pulses, and the output is the final predicted value of the aperture after deviation correction.

[0079] In this embodiment, for any number of pulses N, the corresponding final predicted aperture value is... :

[0080]

[0081] This leads to the "LuGre+Kriging" joint prediction model that predicts the relationship between aperture change and input pulse number N.

[0082] S5. Set the target aperture diameter, calculate the required number of rotor input pulses using the joint prediction model, execute the number of rotor input pulses by the drive motor, calculate the aperture diameter measurement value and compare it with the target aperture diameter. If the error is greater than the set threshold, then correct the number of rotor input pulses and update the training dataset to achieve adaptive aperture control.

[0083] In this embodiment, a target aperture is set to achieve closed-loop control of the aperture. The required number of pulses N is calculated using a joint prediction model. The drive motor executes N pulses to update the system state. The rotor angle is read using a hollow shaft optical encoder and directly fed back to the MCU to calculate the actual aperture size. If the error Then, the pulse input is readjusted, the newly acquired data is periodically added to the training set, the Kriging model is updated, the system adapts, and a high-precision aperture control method based on the joint LuGre friction model and Kriging surrogate model is completed.

[0084] Example 2

[0085] This embodiment provides an adaptive aperture control system, including:

[0086] Basic physical model construction module: used to determine the static parameter set and dynamic parameter set of LuGre dynamic friction model through a step-by-step identification strategy. The rotor input pulse number of the drive motor that controls the aperture is used as the input of LuGre dynamic friction model, and the aperture prediction value is used as the output of LuGre dynamic friction model to construct the basic model for aperture prediction.

[0087] Data acquisition module: used to collect the rotor input pulse count of the control aperture and the corresponding aperture diameter measurement value to build a training dataset;

[0088] Error compensator construction module: It is used to input the training dataset into the aperture prediction base model, obtain the deviation value between the aperture prediction value and the aperture measurement value corresponding to the same rotor input pulse number, and use the deviation value to train the error compensator based on the Kriging surrogate model;

[0089] Joint prediction model construction module: used to combine the basic model of aperture prediction with the error compensator to build a joint prediction model. The input value of the joint prediction model is the number of rotor input pulses, and the output is the final predicted value of aperture after deviation correction.

[0090] Aperture closed-loop control module: Used to set the target aperture diameter, calculate the required number of rotor input pulses using a joint prediction model, drive the motor to execute the number of rotor input pulses, calculate the aperture diameter measurement value and compare it with the target aperture diameter. If the error is greater than the set threshold, the number of rotor input pulses is corrected and the training dataset is updated to achieve adaptive aperture control.

[0091] Example 3

[0092] To verify the effectiveness of the method described in Example 1 and the system described in Example 2, this example is based on experimental verification and explanation in a real-world application scenario.

[0093] For the convenience of experimental research, an existing piezoelectric aperture mechanism was selected. The specific structure of this mechanism is exactly the same as that described in Chinese invention patent CN120595409A, "Piezoelectric Variable Optical Aperture Aperture," and will not be repeated here. The control system components of this piezoelectric aperture mechanism include: an ultrasonic motor and its drive circuit for receiving pulse signals and driving the aperture mechanism; an aperture size detection module for real-time extraction of aperture values; a main control unit using STM32 and FPGA, responsible for signal generation, data acquisition, model calculation, and control decisions; a storage module for storing LuGre model parameters, Kriging model hyperparameters, and historical data; and a communication interface supporting host computer debugging and external command input. A high-precision hollow shaft optical encoder is directly installed inside the rotor of the aperture. The encoder's output line is connected to a specific pin of the MCU, and the MCU determines the rotation angle and direction by reading the number of pulses and the phase difference.

[0094] Parameter identification was performed first: a constant-speed experiment was conducted, inputting pulse signals of 10–500 Hz in 10 Hz increments, and recording the steady-state aperture. Then, a dynamic experiment was conducted, inputting sinusoidal pulse waves of 0.1–5 Hz frequency. After data acquisition, the formulas were solved using the fmincon optimization toolbox in MATLAB. and The LuGre model parameters were obtained. Then, the elastic deformation variable z of the bristles was continuously updated to improve the identification of dynamic friction parameters.

[0095] Training the Kriging model: Gaussian process regression is implemented using Python's scikit-learn library. The RBF kernel function is selected to train and predict the error sequence.

[0096] Control closed-loop implementation: A joint prediction algorithm is embedded in the MCU, with a control cycle of 10 ms. Every 100 control actions are completed, new data is stored in a circular buffer. If the data update rate exceeds 5%, the Kriging model online update program is triggered.

[0097] The main controller (MCU) runs the LuGre+Kriging model to calculate the number of pulses N to be executed and outputs it to the high-voltage drive circuit. Using a full-bridge inverter circuit and a low-pass filter, and employing PWM control technology, the modulation ratio of the modulating wave to the carrier and the transfer function of the low-pass filter are calculated based on the required AC voltage frequency. This AC voltage at that frequency is then applied to the piezoelectric ceramic. Utilizing the inverse piezoelectric effect and frictional coupling, the microscopic vibration of the stator is converted into the macroscopic rotation of the rotor, thereby driving the blades to move synchronously and changing the overlap of the blades, thus achieving dynamic adjustment of the aperture size. The specific principle is the same as the driving method and the principle of converting the microscopic vibration of the stator into the macroscopic rotation of the rotor proposed in the paper "Design of High-Precision Aperture Aperture Driven by Single-Phase Piezoelectric Motor" (Cao Teng, Li Xiaoniu, Wang Baiquan, et al.; Chinese Mechanical Engineering, Vol. 33, No. 20), and will not be elaborated further here.

[0098] The above experimental research is universal. The same beneficial effects can still be obtained by using the control method of this invention for different types of piezoelectric apertures.

[0099] Please see Figure 2 As shown, Figure 2 Table 1 presents a comparison of the step responses obtained from simulations of the aperture adaptive control method provided by this invention and the traditional aperture PID control scheme in this embodiment. The table shows the performance data obtained from the step response comparison. As can be seen from Table 1, this invention has significant advantages over the traditional PID scheme in terms of steady-state error, response time, temperature drift, overshoot, load disturbance, robustness level, and production yield, and has reached a leading level.

[0100] Table 1 Performance indicators This invention Traditional PID Steady-state error (μm) 0.07 0.25 Response time (ms) 48 210 Temperature drift (μm) ±0.04 ±0.31 Overshoot (%) 0 56.4 Load disturbance (μm) ±0.03 ±0.25 Robustness rating Grade A Class C Production yield (%) 99.2 95.8

[0101] Please see Figure 3 As shown, Figure 3 Table 2 shows the predicted surface plot obtained from the Kriging surrogate model. It contains the specific data of the feature points of the surface plot, including the highest point, lowest point, highest confidence point, and best operating point. Figure 3 Region ① is the Stribeck effect surface, region ② is the pre-slip creep surface, and region ③ is the high-confidence surface. Figure 3 Table 2 shows a strongly nonlinear mapping relationship between pulse number, friction torque, and orifice diameter.

[0102] Table 2 Key points Pulse count (P) Torque (τ) Aperture variation (∆D) Confidence highest point 5000 1.5000 20.9108 0.004296 Lowest point 50.5 0.0570 -0.4492 0.000005 highest confidence point 3282.8 0.7025 9.9837 0.999680 Best working point 3434.3 0.7405 10.5345 0.975689

[0103] Figure 3The study demonstrates that the Kriging surrogate model, through the flexible selection of the covariance function, can adaptively model the complex functional relationship between input variables (pulse number P, friction torque τ) and output response (aperture change ∆D), including the synergistic representation of local features and global trends. This achieves an accurate representation of the spatial structure of regionalized variables and ensures that the prediction variance is minimized under the condition of satisfying the unbiasedness constraint, thus achieving the optimal estimation in a statistical sense.

Claims

1. An adaptive aperture control method, characterized in that, Includes the following steps: S1. Determine the static and dynamic parameter sets of the LuGre dynamic friction model through a step-by-step identification strategy. Use the rotor input pulse number of the drive motor that controls the aperture as the input of the LuGre dynamic friction model and the aperture prediction value as the output of the LuGre dynamic friction model to construct the aperture prediction basic model. S2. Collect the rotor input pulse count of the control aperture and the corresponding aperture diameter measurement value to construct a training dataset; S3. Input the training dataset into the aperture prediction base model to obtain the deviation between the aperture prediction value and the aperture measurement value corresponding to the same rotor input pulse number. Use the deviation value to train an error compensator based on the Kriging surrogate model. S4. Combine the basic model for aperture prediction with the error compensator to construct a joint prediction model. The input value of the joint prediction model is the number of rotor input pulses, and the output is the final predicted value of the aperture after deviation correction. S5. Set the target aperture diameter, calculate the required number of rotor input pulses using the joint prediction model, execute the number of rotor input pulses by the drive motor, calculate the aperture diameter measurement value and compare it with the target aperture diameter. If the error is greater than the set threshold, then correct the number of rotor input pulses and update the training dataset to achieve adaptive aperture control.

2. The aperture adaptive control method according to claim 1, characterized in that, The step-by-step identification strategy for determining the static and dynamic parameter sets of the LuGre dynamic friction model includes: identifying static parameters based on constant-speed experimental data, identifying dynamic parameters based on variable-speed experimental data and the static parameter set, improving the dynamic parameter identification by updating the bristle elastic deformation, and finally determining the static and dynamic parameter sets.

3. The aperture adaptive control method according to claim 2, characterized in that, The identification of static parameters based on constant-rate experimental data includes: Data on the aperture of the drive motor at different constant speeds are collected. Based on the Stribeck friction model, output parameters corresponding to each combination of static parameters and the speed of the drive motor are constructed. The output parameters are the difference between the actual steady-state friction force and the calculated steady-state friction force. The absolute value of the output parameters is minimized through a global optimization algorithm to obtain the corresponding static parameters. The static parameters include Coulomb friction force, maximum static friction coefficient, Stribeck characteristic velocity, and viscous friction coefficient.

4. The aperture adaptive control method according to claim 3, characterized in that, The Stribeck friction model formula is as follows: In the formula: For steady-state friction, Coulomb friction, For maximum static friction, For relative velocity, Stribeck characteristic velocity, is the coefficient of viscous friction.

5. The aperture adaptive control method according to claim 3, characterized in that, The identification of dynamic parameters based on variable speed experimental data and static parameter sets includes: based on the obtained static parameters, collecting aperture data during the variable speed motion of the drive motor; constructing the output parameters corresponding to each dynamic parameter combination at a certain moment during the variable speed motion based on the dynamic LuGre friction model; the output parameters are the difference between the actual value of dynamic friction and the calculated value of dynamic friction; minimizing the absolute value of the output parameters through a global optimization algorithm to obtain the corresponding dynamic parameters, which include bristle stiffness and bristle damping coefficient.

6. The aperture adaptive control method according to claim 5, characterized in that, The dynamic LuGre friction force model formula is as follows: In the formula: For dynamic friction, For bristle stiffness, This is the bristle damping coefficient. The coefficient of viscous friction is... For the deformation of the mane, This refers to relative velocity.

7. The aperture adaptive control method according to claim 6, characterized in that, The step of improving dynamic parameter identification by updating the elastic deformation of the mane includes: designing a nonlinear observer, as shown in the following formula: In the formula: This is an estimate of the mane deformation. Based on The predicted value of friction is obtained. This is the measured value of friction force. For observer gain; This represents the steady-state deformation. By embedding the nonlinear observer into the dynamic parameter identification process, the dynamic parameters can be improved.

8. The aperture adaptive control method according to claim 1, characterized in that, The formula for the final predicted value of the aperture is as follows: In the formula: This is the final predicted value for the aperture. The predicted aperture value of the grating is obtained from the basic model for aperture prediction. This is the deviation value obtained from the Kriging surrogate model.

9. The aperture adaptive control method according to claim 1, characterized in that, Step S2 includes: connecting the rotor of the drive motor that controls the aperture diameter to a hollow shaft optical encoder, and using the hollow shaft optical encoder to collect the rotor input pulse number and the corresponding aperture diameter measurement value.

10. An adaptive aperture control system, characterized in that, include: Basic physical model construction module: used to determine the static parameter set and dynamic parameter set of LuGre dynamic friction model through a step-by-step identification strategy. The rotor input pulse number of the drive motor that controls the aperture is used as the input of LuGre dynamic friction model, and the aperture prediction value is used as the output of LuGre dynamic friction model to construct the basic model for aperture prediction. Data acquisition module: used to collect the rotor input pulse count of the control aperture and the corresponding aperture diameter measurement value to build a training dataset; Error compensator construction module: It is used to input the training dataset into the aperture prediction base model, obtain the deviation value between the aperture prediction value and the aperture measurement value corresponding to the same rotor input pulse number, and use the deviation value to train the error compensator based on the Kriging surrogate model; Joint prediction model construction module: used to combine the basic model of aperture prediction with the error compensator to build a joint prediction model. The input value of the joint prediction model is the number of rotor input pulses, and the output is the final predicted value of aperture after deviation correction. Aperture closed-loop control module: Used to set the target aperture diameter, calculate the required number of rotor input pulses using a joint prediction model, drive the motor to execute the number of rotor input pulses, calculate the aperture diameter measurement value and compare it with the target aperture diameter. If the error is greater than the set threshold, the number of rotor input pulses is corrected and the training dataset is updated to achieve adaptive aperture control.

Images