Ftir moving mirror active disturbance rejection control parameter collaborative optimization method and device based on joint bayesian optimization

Abstract

The application provides an FTIR moving mirror active disturbance rejection control parameter collaborative optimization method and device based on joint Bayesian optimization, which comprises the following steps: establishing an ADRC control state equation for an FTIR moving mirror control system, and an ADRC controller comprising a plurality of parameters to be tuned; presetting standard working condition reference input and complex working condition reference input; simulating the standard working condition reference input and the complex working condition reference input based on the ADRC control state equation, and constructing a comprehensive target optimization function; and performing joint Bayesian optimization based on the comprehensive target optimization function. In the scheme, the comprehensive target optimization function is constructed through double-working-condition simulation of the preset standard working condition and the complex working condition, so as to fully cover real complex conditions such as disturbance amplitude change and command speed switching, avoid performance degradation caused by single-working-condition optimization, and obtain better optimization parameters.

Description

Technical Field

[0001] This application relates to the field of precision motion control, and in particular to a method and apparatus for collaborative optimization of FTIR moving mirror active disturbance rejection control parameters based on joint Bayesian optimization. Background Technology

[0002] Fourier transform infrared (FTIR) spectrometers, with their advantages of wide spectral range, high signal-to-noise ratio, and fast measurement speed, are widely used in environmental monitoring, material composition analysis, and biomedical testing. Their core working principle relies on the reciprocating uniform scanning of a moving mirror in a Michelson interferometer. By continuously changing the optical path difference, interference signals are acquired, and finally, the target spectrum is obtained through Fourier transform. The uniformity of the moving mirror's motion speed and the accuracy of its trajectory tracking directly determine the accuracy of the optical path difference sampling interval, such as the interferogram, and are core factors affecting the final spectral resolution, signal-to-noise ratio, and wavelength calibration accuracy. In practical engineering scenarios, voice coil motor-driven moving mirror systems generally face multiple uncertainties, including mechanical damping fluctuations, load disturbances, sensor measurement noise, and environmental temperature drift, placing stringent requirements on the control system's response speed, steady-state accuracy, and disturbance rejection robustness.

[0003] Early FTIR moving mirror control systems mostly adopted classic PID control schemes. These schemes achieved large-scale implementation due to their simple structure, clear parameter physical meaning, and ease of engineering implementation. Later, improved forms such as fuzzy PID and rule-based self-learning PID were developed, which can meet basic control requirements in relatively stable operating conditions. However, PID control is essentially error-driven feedback control, lacking active observation and feedforward compensation mechanisms for unmodeled internal dynamics and external load disturbances. Under complex operating conditions with strong disturbances and fast responses, it is difficult to simultaneously achieve overshoot suppression, adjustment speed, and steady-state accuracy, gradually becoming a bottleneck restricting further improvement in FTIR spectral performance. Active disturbance rejection control (ADRC), through an extended state observer (ESO), uniformly classifies internal and external disturbances into a "total disturbance" for real-time estimation and compensation, possessing stronger disturbance rejection capability and robustness. It has been gradually applied in the field of high-precision voice coil motor motion control, providing a better technical path for FTIR moving mirror control.

[0004] However, ADRC controllers have high parameter dimensionality and strong coupling, including multiple adjustable parameters such as controller bandwidth, nonlinear range width, and observer bandwidth. Traditional manual tuning relies heavily on engineering experience, has a long debugging cycle, and is difficult to achieve globally optimal performance.

[0005] In specific applications of FTIR moving mirror control, existing ADRC parameter optimization methods still have two significant technical shortcomings. First, insufficient coverage of operating conditions: most studies only focus on parameter optimization under two simple operating conditions—"no disturbance" and "disturbance"—failing to fully consider complex conditions such as changes in disturbance amplitude, measurement noise coupling, and reference command speed switching in actual operation. This leads to performance degradation of parameters optimized under a single operating condition in multi-condition environments, resulting in insufficient generalization ability. Second, incomplete parameter optimization scope: existing schemes generally only optimize the parameters of the ADRC master controller, while the core parameters of the extended state observer (ESO) always use empirical preset values, ignoring the strong coupling relationship between the controller and the observer—the disturbance estimation accuracy of the ESO directly affects the compensation effect of the control quantity, and changes in controller bandwidth also change the optimal configuration of the observer. Separate tuning cannot unleash the performance potential of the controller and observer working together, especially in complex operating conditions containing both noise and fast reference inputs, where empirical ESO parameters often become the bottleneck of system performance. Summary of the Invention

[0006] This application provides a method and apparatus for collaborative optimization of FTIR moving mirror active disturbance rejection control parameters based on joint Bayesian optimization. By constructing a comprehensive objective optimization function through dual-condition simulation of preset standard and complex operating conditions, it can fully cover real and complex conditions such as disturbance amplitude changes and command speed switching, avoid performance degradation caused by single-condition optimization, and thus obtain better optimization parameters.

[0007] In a first aspect, embodiments of this application provide a collaborative optimization method for FTIR moving mirror active disturbance rejection control parameters based on joint Bayesian optimization, the method comprising: Taking the FTIR moving mirror control system as an example, the ADRC control state equation is established, and an ADRC controller including multiple parameters to be tuned is constructed. The ADRC controller simulates the standard operating condition reference input without noise disturbance and the complex operating condition reference input with noise disturbance based on the ADRC control state equation. The simulation results are obtained as follows: standard performance index under standard operating condition and complex performance index under complex operating condition. The combined objective optimization function is a weighted sum of standard performance indicators and complex performance indicators. The combined objective optimization function is used as the optimization objective to iteratively perform joint Bayesian optimization on each parameter to be tuned until the preset optimization objective is met. The optimal parameter set is formed by obtaining the optimal value of each parameter to be tuned. The ADRC controller is then configured with the optimal parameter set. The weight of the complex performance indicators is greater than the weight of the standard performance indicators.

[0008] Secondly, embodiments of this application provide a collaborative optimization device for FTIR moving mirror active disturbance rejection control parameters based on joint Bayesian optimization, comprising: The module is constructed to establish the ADRC control state equations for the FTIR moving mirror control system and to construct an ADRC controller with multiple parameters to be tuned. The simulation module is preset with a standard operating condition reference input that does not include noise disturbance and a complex operating condition reference input that includes noise disturbance. The ADRC controller simulates the standard operating condition reference input and the complex operating condition reference input based on the ADRC control state equation, respectively, and obtains the standard performance index under the standard operating condition and the complex performance index under the complex operating condition. The iterative optimization module uses the weighted sum of standard performance indicators and complex performance indicators as the comprehensive objective optimization function. It uses the comprehensive objective optimization function as the optimization objective to iteratively perform joint Bayesian optimization on each parameter to be tuned until the preset optimization objective is met. The optimal value of each parameter to be tuned is obtained to form the optimal parameter set. The ADRC controller is configured with the optimal parameter set. The weight of the complex performance indicators is greater than the weight of the standard performance indicators.

[0009] Thirdly, embodiments of this application provide an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute a collaborative optimization method for FTIR moving mirror active disturbance rejection control parameters based on joint Bayesian optimization.

[0010] Fourthly, embodiments of this application provide a readable storage medium storing a computer program, which, when executed by a processor, implements a collaborative optimization method for FTIR moving mirror active disturbance rejection control parameters based on joint Bayesian optimization.

[0011] The main contributions and innovations of this invention are as follows: This application embodiment constructs a multi-parameter ADRC controller containing a sigmoid reference trajectory shaping module, an extended state observer, and a nonlinear state error feedback control law. The sigmoid reference trajectory shaping module suppresses the shock and overshoot caused by sudden changes in the reference signal by smoothing the target reference signal. The extended state observer unifies the system's internal uncertainties and external disturbances into a total disturbance, observes and compensates for it in real time, significantly improving the disturbance rejection capability and robustness. The nonlinear feedback suppresses overshoot while taking into account response speed and steady-state accuracy. This application embodiment sets a standard operating condition reference input without noise disturbance and a complex operating condition reference input containing disturbance, measurement noise, and fast reference input. The ADRC controller simulates and collects performance indicators under both types of operating conditions, fully covering real and complex conditions such as disturbance amplitude changes, measurement noise coupling, and command speed switching, avoiding performance degradation caused by single-condition optimization. This significantly enhances the generalization ability of parameters under multiple operating conditions. The embodiments of this application use a weighted sum of standard performance indicators and complex performance indicators as the comprehensive objective, giving a higher weight to complex performance indicators than standard performance indicators. Settlement time is given a higher weight in both types of operating conditions. While balancing response speed and steady-state accuracy, it also prioritizes the robustness against disturbances in complex operating conditions, ensuring the optimization results focus more on the critical disturbance resistance performance in practical engineering while maintaining rapid response. The embodiments of this application first jointly optimize the main control parameters such as transition time constant, transition time gain, controller bandwidth, and control nonlinear interval width. Then, the optimal value of the main control parameter is fixed, and a neighborhood search algorithm is used to optimize the observation parameters such as observer bandwidth and observation nonlinear interval width. Finally, a search space is constructed in the neighborhood centered on the optimal value of each parameter, and all parameters are jointly optimized again. This breaks through the limitations of traditional methods that only optimize the main control parameters while using empirically preset values ​​for the core ESO parameters. This phased strategy enhances the disturbance observation and compensation capabilities under complex operating conditions without compromising the existing performance of the main controller, and provides high-quality initial values ​​of observation parameters for the final joint optimization, thereby releasing the performance potential of the controller and observer working together and avoiding the ESO empirical parameters becoming a performance bottleneck under complex operating conditions; neighborhood search also narrows the search space and further improves optimization efficiency.

[0012] Details of one or more embodiments of this application are set forth in the following drawings and description to make other features, objects and advantages of this application more readily apparent. Attached Figure Description

[0013] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1This is a flowchart illustrating a collaborative optimization method for FTIR moving mirror active disturbance rejection control parameters based on joint Bayesian optimization, according to an embodiment of this application. Figure 2 This is a schematic diagram of the structure of a oscillating sweeping corner mirror interferometer according to an embodiment of this application; Figure 3 This is an overall structural block diagram of an ADRC controller according to an embodiment of this application; Figure 4 This is a schematic diagram of a process for iteratively performing joint Bayesian optimization on each parameter to be tuned according to an embodiment of this application; Figure 5 This is a bar chart comparing the performance of this solution with that of a manually adjusted ADRC baseline under standard and complex operating conditions. Figure 6 This is a structural block diagram of a collaborative optimization device for FTIR moving mirror active disturbance rejection control parameters based on joint Bayesian optimization, according to an embodiment of this application. Figure 7 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of this application. Detailed Implementation

[0014] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with one or more embodiments of this specification. Rather, they are merely examples of apparatuses and methods consistent with some aspects of one or more embodiments of this specification as detailed in the appended claims.

[0015] It should be noted that the steps of the corresponding methods are not necessarily performed in the order shown and described in this specification in other embodiments. In some other embodiments, the methods may include more or fewer steps than described in this specification. Furthermore, a single step described in this specification may be broken down into multiple steps in other embodiments; and multiple steps described in this specification may be combined into a single step in other embodiments.

[0016] Example 1 This application provides a collaborative optimization method for FTIR moving mirror active disturbance rejection control parameters based on joint Bayesian optimization. It constructs a comprehensive objective optimization function through dual-condition simulation of preset standard and complex operating conditions to fully cover real-world complex conditions such as disturbance amplitude variations and command speed switching, avoiding performance degradation caused by single-condition optimization, thereby obtaining better optimization parameters. Specifically, refer to... Figure 1 The method includes: Taking the FTIR moving mirror control system as an example, the ADRC control state equation is established, and an ADRC controller including multiple parameters to be tuned is constructed. The ADRC controller simulates the standard operating condition reference input without noise disturbance and the complex operating condition reference input with noise disturbance based on the ADRC control state equation. The simulation results are obtained as follows: standard performance index under standard operating condition and complex performance index under complex operating condition. The combined objective optimization function is a weighted sum of standard performance indicators and complex performance indicators. The combined objective optimization function is used as the optimization objective to iteratively perform joint Bayesian optimization on each parameter to be tuned until the preset optimization objective is met. The optimal parameter set is formed by obtaining the optimal value of each parameter to be tuned. The ADRC controller is then configured with the optimal parameter set. The weight of the complex performance indicators is greater than the weight of the standard performance indicators.

[0017] In the current embodiment, this scheme uses a pendulum sweeping corner mirror interferometer as the controlled object. Voltage balance equations and force balance equations for the voice coil motor driving the moving mirror are established for the controlled object. The voltage balance equations and force balance equations are transformed by Laplace and reduced in order to obtain a second-order transfer function. Based on the second-order transfer function, an ADRC control state equation containing displacement state variables, velocity state variables, total disturbance and system gain is established.

[0018] Specifically, the structure of the oscillating sweeping corner mirror interferometer is as follows: Figure 2 As shown, in this scheme, starting from the physical equations, the actual moving mirror system of the pendulum sweeping corner mirror interferometer is transformed into a mathematically simulable object. The voltage balance equation is expressed as follows:

[0019] The force balance equation is expressed as follows:

[0020] in, The voltage across the coil terminals. For equivalent inductance, Equivalent resistance For current, The equivalent length of the coil, The magnetic field strength, For load quality, The damping coefficient is... The fulcrum is the equivalent drag coefficient. The angle of the moving mirror's swing. This represents the displacement of the moving mirror line. For time, then For current The differential, For time The differential, For the displacement of the moving mirror line The differential, Let be the derivative of current with respect to time. Let be the derivative of the displacement of the moving mirror line with respect to time. Let be the second derivative of the displacement of the moving mirror line with respect to time.

[0021] Specifically, the optical path difference With the angle of the moving mirror The relationship is The rate of change of optical path difference is ,in Given the length of the swing arm, the second-order transfer function obtained after Laplace transform and order reduction is... for:

[0022] Furthermore, the ADRC control state equation is used to control the oscillating corner mirror interferometer using a mathematical model, wherein the displacement state variable is the linear displacement of the moving mirror; the velocity state variable is the velocity of the moving mirror; the total disturbance is a comprehensive disturbance term including internal system uncertainties and external disturbances; and the system gain is the inherent control gain derived from the physical model of the controlled object.

[0023] In the current embodiment, the ADRC controller includes a sigmoid reference trajectory shaping module, an extended state observer, and a nonlinear state error feedback control law. The sigmoid reference trajectory shaping module is used to smooth and shape the target reference signal to obtain a shaped reference trajectory. The extended state observer acquires the displacement state estimate, velocity state estimate, and total disturbance estimate of the FTIR moving mirror control system in real time. The nonlinear state error feedback control law generates intermediate control quantities based on position error and velocity error, and then uses the total disturbance estimate to compensate for the intermediate control quantities to obtain the final control quantity. The final control quantity is then input into the ADRC control state equation for simulation. The position error is obtained based on the shaped reference trajectory and displacement state estimate, and the velocity error is obtained based on the shaped reference trajectory and velocity state estimate. The target reference signal is a standard operating condition reference input or a complex operating condition reference input.

[0024] Furthermore, the parameters to be tuned for the sigmoid reference trajectory shaping module include the transition time constant and the transition time gain; the parameters to be tuned for the extended state observer include the observer bandwidth and the width of the observation nonlinear interval; and the parameters to be tuned for the nonlinear state error feedback control law include the controller bandwidth and the width of the control nonlinear interval.

[0025] Specifically, in the sigmoid reference trajectory shaping module, the transition time constant is used to control the transition speed of the target reference signal's trajectory from the initial value to the steady-state value. That is, the larger the transition time constant, the smoother the transition of the target reference signal's trajectory.

[0026] Specifically, the shaped reference trajectory generated by the sigmoid reference trajectory shaping module includes a reference position and a reference velocity. The reference position is obtained by smoothing and shaping the target position in the target reference signal. The reference velocity is the first derivative of the reference position with respect to time. In other words, the position error is the difference between the reference position and the displacement state estimate, and the velocity error is the difference between the reference velocity and the velocity state estimate.

[0027] Furthermore, the extended state observer is a third-order nonlinear structure. In the extended state observer, the displacement gain coefficient, velocity gain coefficient, and total disturbance estimation coefficient are obtained by parameterizing the observer bandwidth. The displacement state estimate, velocity state estimate, and total disturbance estimate are calculated based on the displacement gain coefficient, velocity gain coefficient, and total disturbance estimation coefficient using nonlinear functions. The width of the observation nonlinear interval is used to limit the activation interval of the nonlinear function. When the observation error is within the nonlinear interval, nonlinear gain adjustment is used. When the observation error exceeds the nonlinear interval, linear gain adjustment is used. The observation error is the difference between the corresponding displacement state estimate and the actual displacement measurement value of the FTIR moving mirror control system.

[0028] Specifically, the formula for parameterizing the observer bandwidth is expressed as follows: , ,

[0029] in, For the observer bandwidth, This is the displacement gain coefficient. This is the velocity gain coefficient. These are the estimated coefficients for the total disturbance.

[0030] Specifically, the formulas for calculating the displacement state estimate, velocity state estimate, and total disturbance estimate are expressed as follows:

[0031]

[0032]

[0033]

[0034] in, For time step number, For observation error, This represents the actual displacement measurement value of the FTIR moving mirror control system. For the corresponding displacement state estimation, For displacement state estimation, For velocity state estimation, For the total disturbance estimate, This is the displacement gain coefficient. This is the velocity gain coefficient. The total disturbance estimation coefficients are... To observe the width of the nonlinear interval, It is a nonlinear function. for Control variables for time steps For system gain, The step size of the time step. , It is a constant in this scheme. .

[0035] In other words, when the observation error is less than or equal to the width of the observation nonlinear interval, i.e. hour, Nonlinear gain adjustment is employed; however, when the observation error exceeds the width of the observation nonlinear interval, i.e. hour, Linear gain adjustment is employed.

[0036] Furthermore, in the nonlinear state error feedback control law, the controller bandwidth and system gain are parameterized to obtain integral, proportional, and derivative channels. The intermediate control quantity is then calculated using nonlinear functions based on the integral, proportional, and derivative channels, position error, and velocity error.

[0037] Specifically, the formulas for parameterizing the controller bandwidth and system gain are expressed as follows: , ,

[0038] in, For integration, As a proportion, For differential channels, For controller bandwidth, This represents the system gain.

[0039] The formula for obtaining the intermediate control quantity is expressed as:

[0040] in, This is an intermediate control variable. For integration, As a proportion, For differential channels, It is a nonlinear function. For positional error, For speed error, To control the width of the nonlinear interval, for The points, for The points, , It is a constant in this scheme. .

[0041] The formula for compensating the intermediate control quantity for disturbance using the total disturbance estimate is expressed as follows:

[0042] in, For the final control quantity, This is an intermediate control variable. For the total disturbance estimate, This represents the system gain.

[0043] In this scheme, the overall structure block diagram of the ADRC controller is as follows: Figure 3 As shown, the target reference signal is first input into the sigmoid reference trajectory shaping module and smoothed to obtain the shaped reference trajectory. The difference between the shaped reference trajectory and the displacement state estimate and velocity state estimate output by the extended state observer is used to obtain the position error and velocity error. The nonlinear state error feedback control law generates intermediate control quantity based on the position error and velocity error. The intermediate control quantity is then subjected to disturbance compensation to obtain the final control quantity. The final control quantity is applied to the controlled object, and the output of the controlled object is fed back to the extended state observer for state observation, thus forming a perfect closed loop.

[0044] In the current embodiment, both the standard performance index and the complex performance index are weighted sums of overshoot, settling time, integral absolute error, and steady-state error, wherein the settling time has a greater weight than the overshoot, integral absolute error, and steady-state error.

[0045] Specifically, this scheme designs two typical operating conditions to uniformly evaluate multiple parameters to be tuned. The standard operating condition reference input is the standard operating condition without disturbance; the complex operating condition parameter input is the complex operating condition that simultaneously includes disturbance, measurement noise and fast reference input.

[0046] Specifically, when obtaining standard performance indicators, the corresponding overshoot, settling time, integral absolute error, and steady-state error are normalized by dividing by their respective baseline reference values ​​and then weighted and summed.

[0047] Similarly, when obtaining complex performance indicators, the overshoot, settling time, integral absolute error, and steady-state error are normalized by dividing by their respective baseline reference values ​​and then summed by weight.

[0048] In the current embodiment, when constructing the comprehensive objective optimization function, the weight of the complex performance index is greater than that of the standard performance index, so as to give more emphasis to robust performance. The settling time is given a high weight in both types of operating conditions to ensure response speed.

[0049] For example, the formula for the standard performance index is expressed as follows:

[0050] in, As standard performance indicators, For the corresponding overshoot, For the corresponding adjustment time, This represents the corresponding integral absolute error. This represents the corresponding steady-state error.

[0051] The formula for the complexity performance index is expressed as:

[0052] in, As a complex performance index, For the corresponding overshoot, For the corresponding adjustment time, This represents the corresponding integral absolute error. This represents the corresponding steady-state error.

[0053] The formula for the comprehensive objective function can then be expressed as:

[0054] in, For the comprehensive objective optimization function, As standard performance indicators, It is a complex performance index.

[0055] In the current embodiment, a flowchart illustrating the iterative joint Bayesian optimization for each parameter to be tuned is shown below. Figure 4 As shown, the steps for iteratively performing joint Bayesian optimization on each parameter to be tuned include: using the transition time constant, transition time gain controller bandwidth, and control nonlinearity interval width as the master control parameters to be tuned, and the observer bandwidth and observation nonlinearity interval width as the observation parameters to be tuned. S1. A probabilistic proxy model using Gaussian process as the comprehensive objective optimization function, with expected improvement criterion as the acquisition function, performs joint optimization on all main control parameters to be tuned under standard working condition reference input and complex working condition reference input, and obtains the main control optimal value of each main control parameter to be tuned. S2. Fix the optimal value of each master control parameter to be tuned obtained from S1, and use the neighborhood search algorithm to optimize each observed parameter to be tuned to obtain the optimal value of each observed parameter to be tuned. S3. For each parameter to be tuned, a neighborhood search is performed centered on the corresponding master optimal value or the observed optimal value to obtain the corresponding search space. A probabilistic surrogate model with a Gaussian process as the comprehensive objective optimization function is adopted. With the expected improvement criterion as the acquisition function, all parameters to be tuned are jointly optimized in the corresponding search space under standard working condition reference input and complex working condition reference input to obtain the optimal value of each parameter to be tuned, forming the optimal parameter set.

[0056] Specifically, existing ADRC controller optimization methods are usually based on one-time settings and lack fine-grained optimization for actual operating conditions. If only the master controller parameters to be tuned are optimized while the observed parameters to be tuned are ignored, it is difficult to fully explore the potential performance of the ADRC controller. Therefore, this scheme adopts a phased optimization approach, first optimizing the master controller parameters to be tuned, then optimizing the observed parameters to be tuned, and finally performing joint optimization based on the optimization results. This enhances the disturbance observation and compensation capabilities under complex operating conditions without compromising the existing performance of the master controller, and at the same time provides high-quality initial values ​​for the observed parameters to be tuned for subsequent joint optimization, which is different from the existing practice of setting the observed parameters to be tuned one-time based on experience.

[0057] Specifically, when performing a neighborhood search in S3, a search space is constructed within the 30% neighborhood of the corresponding parameter.

[0058] Specifically, the preset optimization objective in this scheme is for the comprehensive objective optimization function to converge or reach a preset number of iterations. In each iteration, the next set of parameters to be evaluated is determined to maximize the expected improvement. The six parameters are substituted into the ADRC model and the simulation is run under two types of working conditions to calculate the objective function value and update the dataset.

[0059] Specifically, the expected improvement criterion for the Bayesian optimization is:

[0060] in, For the desired improvement value, The minimum observed value of the integrated objective function is obtained. The value of the comprehensive objective function corresponding to the current parameters to be tuned. These are the parameters to be evaluated.

[0061] Specifically, after obtaining the six optimal parameters to be tuned—transition time constant, transition time gain, observer bandwidth, observation nonlinear interval width, controller bandwidth, and control nonlinear interval width—the displacement gain coefficient, velocity gain coefficient, and total disturbance estimation coefficient are calculated based on the optimal observer bandwidth. The ADRC controller is then configured with the optimal transition time constant, transition time gain, observer bandwidth, observation nonlinear interval width, controller bandwidth, control nonlinear interval width, displacement gain coefficient, velocity gain coefficient, and total disturbance estimation coefficient.

[0062] In some embodiments, compared to a manually adjusted ADRC baseline, the performance indicators of this solution under standard and complex operating conditions are as follows: Figure 5 As shown in Table 1, the quantitative improvement results under standard and complex operating conditions are as follows: Table 1 shows the quantitative improvement results of the optimized parameters and ADRC baseline under standard and complex operating conditions.

[0063] Table 1 shows that under standard operating conditions, the overshoot improved by 94.07% (from 0.1584 to 0.0094), and the steady-state error improved by 94.46%. Under complex operating conditions, the overshoot improved by 91.08%, the steady-state error improved by 94.48%, and the settling time improved by 26.16% (from 0.474s to 0.350s). All indicators showed positive improvement with no degradation terms. The overall objective function decreased from 0.569 to 0.270 (an improvement of 52.50%), verifying the effectiveness of the joint optimization method.

[0064] Example 2 Based on the same concept, referencing Figure 6 This application also proposes a collaborative optimization device for the active disturbance rejection control parameters of FTIR moving mirrors based on joint Bayesian optimization, comprising: The module is constructed to establish the ADRC control state equations for the FTIR moving mirror control system and to construct an ADRC controller with multiple parameters to be tuned. The simulation module is preset with a standard operating condition reference input that does not include noise disturbance and a complex operating condition reference input that includes noise disturbance. The ADRC controller simulates the standard operating condition reference input and the complex operating condition reference input based on the ADRC control state equation, respectively, and obtains the standard performance index under the standard operating condition and the complex performance index under the complex operating condition. The iterative optimization module uses the weighted sum of standard performance indicators and complex performance indicators as the comprehensive objective optimization function. It uses the comprehensive objective optimization function as the optimization objective to iteratively perform joint Bayesian optimization on each parameter to be tuned until the preset optimization objective is met. The optimal value of each parameter to be tuned is obtained to form the optimal parameter set. The ADRC controller is configured with the optimal parameter set. The weight of the complex performance indicators is greater than the weight of the standard performance indicators.

[0065] Example 3 This embodiment also provides an electronic device, see reference. Figure 7 It includes a memory 402 and a processor 401, the memory 402 storing a computer program and the processor 401 being configured to run the computer program to perform the steps in any of the above method embodiments.

[0066] Specifically, the processor 401 may include a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0067] The memory 402 may include a mass storage device for data or instructions. For example, and not limitingly, the memory 402 may include a hard disk drive (HDD), a floppy disk drive, a solid-state drive (SSD), flash memory, an optical disk drive, a magneto-optical disk drive, magnetic tape, or a Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory 402 may include removable or non-removable (or fixed) media. Where appropriate, the memory 402 may be internal or external to a data processing device. In a particular embodiment, the memory 402 is non-volatile memory. In a particular embodiment, the memory 402 includes read-only memory (ROM) and random access memory (RAM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable read-only memory (PROM), an erasable read-only memory (EPROM), an electrically erasable read-only memory (EEPROM), an electrically alterable read-only memory (EAROM), or flash memory, or a combination of two or more of these. Where appropriate, the RAM can be Static Random-Access Memory (SRAM) or Dynamic Random-Access Memory (DRAM). DRAM can be Fast Page Mode Dynamic Random Access Memory (FPMDRAM), Extended Data Out Dynamic Random Access Memory (EDODRAM), Synchronous Dynamic Random-Access Memory (SDRAM), etc.

[0068] The memory 402 can be used to store or cache various data files that need to be processed and / or communicated, as well as possible computer program instructions executed by the processor 401.

[0069] The processor 401 reads and executes the computer program instructions stored in the memory 402 to implement any of the FTIR moving mirror active disturbance rejection control parameter co-optimization methods based on joint Bayesian optimization in the above embodiments.

[0070] Optionally, the electronic device may further include a transmission device 403 and an input / output device 404, wherein the transmission device 403 is connected to the processor 401 and the input / output device 404 is connected to the processor 401.

[0071] The transmission device 403 can be used to receive or send data via a network. Specific examples of the network described above may include wired or wireless networks provided by the communication provider of the electronic device. In one example, the transmission device includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device 403 may be a Radio Frequency (RF) module used for wireless communication with the Internet.

[0072] Input / output device 404 is used to input or output information. In this embodiment, the input information may be multiple initial parameters to be tuned, and the output information may be an optimal set of parameters.

[0073] Optionally, in this embodiment, the processor 401 can be configured to perform the following steps via a computer program: Taking the FTIR moving mirror control system as an example, the ADRC control state equation is established, and an ADRC controller including multiple parameters to be tuned is constructed. The ADRC controller simulates the standard operating condition reference input without noise disturbance and the complex operating condition reference input with noise disturbance based on the ADRC control state equation. The simulation results are obtained as follows: standard performance index under standard operating condition and complex performance index under complex operating condition. The combined objective optimization function is a weighted sum of standard performance indicators and complex performance indicators. The combined objective optimization function is used as the optimization objective to iteratively perform joint Bayesian optimization on each parameter to be tuned until the preset optimization objective is met. The optimal parameter set is formed by obtaining the optimal value of each parameter to be tuned. The ADRC controller is then configured with the optimal parameter set. The weight of the complex performance indicators is greater than the weight of the standard performance indicators.

[0074] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementations, and will not be repeated here.

[0075] Generally, various embodiments can be implemented in hardware or dedicated circuitry, software, logic, or any combination thereof. Some aspects of the invention can be implemented in hardware, while others can be implemented by firmware or software executed by a controller, microprocessor, or other computing device, but the invention is not limited thereto. Although various aspects of the invention may be shown and described as block diagrams, flowcharts, or using some other graphical representation, it should be understood that, by way of non-limiting example, these blocks, apparatuses, systems, techniques, or methods described herein can be implemented in hardware, software, firmware, dedicated circuitry or logic, general-purpose hardware or controllers or other computing devices, or some combination thereof.

[0076] Embodiments of the present invention can be implemented by computer software, which may be executable by a data processor of a mobile device, such as a processor entity, or by hardware, or by a combination of software and hardware. Computer software or programs (also referred to as program products) including software routines, applets, and / or macros can be stored in any device-readable data storage medium, and they include program instructions for performing specific tasks. The computer program product may include one or more computer-executable components configured to perform the embodiments when the program is run. The one or more computer-executable components may be at least one piece of software code or a portion thereof. Additionally, it should be noted in this respect that, as Figure 7 Any box in the logical flow can represent a program step, or interconnected logic circuits, boxes and functions, or a combination of program steps and logic circuits, boxes and functions. Software can be stored on physical media such as memory chips or blocks of storage implemented within a processor, magnetic media such as hard disks or floppy disks, and optical media such as DVDs and their data variants, CDs, etc. The physical medium is a non-transient medium.

[0077] Those skilled in the art should understand that 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 have been 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.

[0078] The above embodiments are merely illustrative of several implementation methods of this application, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of this application. 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 application should be determined by the appended claims.

Claims

1. A collaborative optimization method for active disturbance rejection control parameters of FTIR moving mirrors based on joint Bayesian optimization, characterized in that, Includes the following steps: Taking the FTIR moving mirror control system as an example, the ADRC control state equation is established, and an ADRC controller including multiple parameters to be tuned is constructed. The ADRC controller simulates the standard operating condition reference input without noise disturbance and the complex operating condition reference input with noise disturbance based on the ADRC control state equation. The simulation results are obtained as follows: standard performance index under standard operating condition and complex performance index under complex operating condition. The combined objective optimization function is a weighted sum of standard performance indicators and complex performance indicators. The combined objective optimization function is used as the optimization objective to iteratively perform joint Bayesian optimization on each parameter to be tuned until the preset optimization objective is met. The optimal parameter set is formed by obtaining the optimal value of each parameter to be tuned. The ADRC controller is then configured with the optimal parameter set. The weight of the complex performance indicators is greater than the weight of the standard performance indicators.

2. The method for collaborative optimization of FTIR moving mirror active disturbance rejection control parameters based on joint Bayesian optimization according to claim 1, characterized in that, This scheme uses a pendulum sweeping corner mirror interferometer as the controlled object. Voltage balance equations and force balance equations for the voice coil motor driving the moving mirror are established for the controlled object. The voltage balance equations and force balance equations are then transformed by Laplace and reduced in order to obtain a second-order transfer function. Based on the second-order transfer function, an ADRC control state equation containing displacement state variables, velocity state variables, total disturbance, and system gain is established.

3. The method for collaborative optimization of FTIR moving mirror active disturbance rejection control parameters based on joint Bayesian optimization according to claim 1, characterized in that, The ADRC controller includes a sigmoid reference trajectory shaping module, an extended state observer, and a nonlinear state error feedback control law. The sigmoid reference trajectory shaping module is used to smooth and shape the target reference signal to obtain a shaped reference trajectory. The extended state observer acquires the displacement state estimate, velocity state estimate, and total disturbance estimate of the FTIR moving mirror control system in real time. The nonlinear state error feedback control law generates intermediate control quantities based on position and velocity errors. Then, the intermediate control quantities are compensated for by the total disturbance estimation to obtain the final control quantity. The final control quantity is then input into the ADRC control state equation for simulation. The position error is obtained based on the shaped reference trajectory and displacement state estimation, and the velocity error is obtained based on the shaped reference trajectory and velocity state estimation. The target reference signal is either a standard working condition reference input or a complex working condition reference input.

4. The method for collaborative optimization of FTIR moving mirror active disturbance rejection control parameters based on joint Bayesian optimization according to claim 3, characterized in that, The parameters to be tuned for the sigmoid reference trajectory shaping module include the transition time constant and the transition time gain; the parameters to be tuned for the extended state observer include the observer bandwidth and the width of the observation nonlinear interval; the parameters to be tuned for the nonlinear state error feedback control law include the controller bandwidth and the width of the control nonlinear interval.

5. The method for collaborative optimization of FTIR moving mirror active disturbance rejection control parameters based on joint Bayesian optimization according to claim 3, characterized in that, The extended state observer is a third-order nonlinear structure. In the extended state observer, the displacement gain coefficient, velocity gain coefficient, and total disturbance estimation coefficient are obtained by parameterizing the observer bandwidth. The displacement state estimate, velocity state estimate, and total disturbance estimate are calculated based on the displacement gain coefficient, velocity gain coefficient, and total disturbance estimation coefficient using nonlinear functions. The width of the observation nonlinear interval is used to limit the activation interval of the nonlinear function. When the observation error is within the nonlinear interval, nonlinear gain adjustment is used. When the observation error exceeds the nonlinear interval, linear gain adjustment is used. The observation error is the difference between the corresponding displacement state estimate and the actual displacement measurement value of the FTIR moving mirror control system.

6. The method for collaborative optimization of FTIR moving mirror active disturbance rejection control parameters based on joint Bayesian optimization according to claim 3, characterized in that, In the nonlinear state error feedback control law, the controller bandwidth and system gain are parameterized to obtain integral, proportional and derivative channels. The intermediate control quantity is calculated based on the integral, proportional and derivative channels, position error and velocity error using nonlinear functions.

7. The method for collaborative optimization of FTIR moving mirror active disturbance rejection control parameters based on joint Bayesian optimization according to claim 4, characterized in that, The steps of iteratively performing joint Bayesian optimization for each parameter to be tuned include: using the transition time constant, transition time gain controller bandwidth, and control nonlinearity interval width as the master control parameters to be tuned, and the observer bandwidth and observation nonlinearity interval width as the observation parameters to be tuned. S1. A probabilistic proxy model using Gaussian process as the comprehensive objective optimization function, with expected improvement criterion as the acquisition function, performs joint optimization on all main control parameters to be tuned under standard working condition reference input and complex working condition reference input, and obtains the main control optimal value of each main control parameter to be tuned. S2. Fix the optimal value of each master control parameter to be tuned obtained from S1, and use the neighborhood search algorithm to optimize each observed parameter to be tuned to obtain the optimal value of each observed parameter to be tuned. S3. For each parameter to be tuned, a neighborhood search is performed centered on the corresponding master optimal value or the observed optimal value to obtain the corresponding search space. A probabilistic surrogate model with a Gaussian process as the comprehensive objective optimization function is adopted. With the expected improvement criterion as the acquisition function, all parameters to be tuned are jointly optimized in the corresponding search space under standard working condition reference input and complex working condition reference input to obtain the optimal value of each parameter to be tuned, forming the optimal parameter set.

8. A collaborative optimization device for FTIR moving mirror active disturbance rejection control parameters based on joint Bayesian optimization, characterized in that, The module is constructed to establish the ADRC control state equations for the FTIR moving mirror control system and to construct an ADRC controller with multiple parameters to be tuned. The simulation module is preset with a standard operating condition reference input that does not include noise disturbance and a complex operating condition reference input that includes noise disturbance. The ADRC controller simulates the standard operating condition reference input and the complex operating condition reference input based on the ADRC control state equation, respectively, and obtains the standard performance index under the standard operating condition and the complex performance index under the complex operating condition. The iterative optimization module uses the weighted sum of standard performance indicators and complex performance indicators as the comprehensive objective optimization function. It uses the comprehensive objective optimization function as the optimization objective to iteratively perform joint Bayesian optimization on each parameter to be tuned until the preset optimization objective is met. The optimal value of each parameter to be tuned is obtained to form the optimal parameter set. The ADRC controller is configured with the optimal parameter set. The weight of the complex performance indicators is greater than the weight of the standard performance indicators.

9. An electronic device comprising a memory and a processor, characterized in that, The memory stores a computer program, and the processor is configured to run the computer program to execute the collaborative optimization method for FTIR moving mirror active disturbance rejection control parameters based on joint Bayesian optimization as described in any one of claims 1-7.

10. A readable storage medium, characterized in that, The readable storage medium stores a computer program, which, when executed by a processor, implements a collaborative optimization method for FTIR moving mirror active disturbance rejection control parameters based on joint Bayesian optimization as described in any one of claims 1-7.

Images