Synchrotron radiation light source micro-vibration high-precision control method based on data driving

Abstract

The invention discloses a synchrotron radiation light source micro-vibration high-precision control method based on data driving, and relates to the technical field of synchrotron radiation light source control, and the method comprises the steps: S1, data collection and preprocessing: collecting various data in the operation process of a synchrotron radiation light source, including but not limited to vibration data, beam energy and an experiment mode; s2, working condition identification and classification; s3, adjusting control strategy parameters; s4, micro-vibration high-precision control is carried out; and S5, performing performance evaluation and feedback optimization. According to the synchrotron radiation light source micro-vibration high-precision control method based on data driving, through data driving, the operation working condition of the light source can be recognized and classified in real time, when the operation working condition of the light source changes, parameters of a control strategy can be automatically adjusted, manual recalibration is not needed, self-adaptive high-precision control over micro-vibration under different working conditions is achieved, and high-precision control over micro-vibration under different working conditions is achieved. And the problem that the control precision is reduced when the working condition is changed in a traditional fixed control strategy is avoided.

Description

Technical Field

[0001] This invention relates to the field of synchrotron radiation source control technology, specifically a data-driven high-precision control method for micro-vibrations of synchrotron radiation sources. Background Technology

[0002] A synchrotron radiation source is a device that utilizes the electromagnetic radiation generated when high-speed electrons move in a circular motion within a magnetic field. When charged particles (such as electrons) move in a circular motion at near the speed of light in a synchrotron accelerator or storage ring, they radiate electromagnetic waves along the tangential direction due to centripetal acceleration. These electromagnetic waves are called synchrotron radiation, and the device that generates this radiation is called a synchrotron radiation source. Synchrotron radiation possesses many excellent characteristics, including a wide spectrum (from infrared to hard X-rays), high brightness, good directionality, high polarization degree, and pulse time structure.

[0003] For example, the patent document CN108222079A discloses a micro-nano-level vibration control technology for the foundation of an ultra-long synchrotron radiation source. This technology employs a pile-raft foundation, a large-volume steel-reinforced concrete ring foundation, and a polyurethane energy-dissipating and vibration-damping layer to improve overall stiffness and reduce the risk of foundation settlement. However, because this technical solution relies on a fixed physical structure for passive vibration reduction, it cannot automatically adjust control parameters according to the operating conditions of the light source. Therefore, when operating conditions change, the control accuracy is prone to decrease.

[0004] For example, the patent document with publication number CN114200758A discloses a light source device, a projection device, a microlens array, and a light source control method. The core of its technical solution lies in miniaturizing the optical system through a microlens array and a focusing optical system. However, it cannot directly and precisely control the micro-vibration or height of the light source, making it unsuitable for application scenarios that require extreme light source stability and precision.

[0005] However, many scientific experiments utilizing synchrotron radiation sources, such as X-ray crystallography analysis and the study of material microstructures, place extremely high demands on the stability and precision of the light source. Even minute vibrations or fluctuations can lead to deviations in experimental data, affecting the accurate analysis of material structures. Furthermore, synchrotron radiation sources face various operating conditions during operation, such as adjustments to beam energy and switching of experimental modes. Different operating conditions place different demands on the stability and control of the light source. Without control, these vibrations will affect the performance of the light source, making it unable to meet the needs of different experiments. Summary of the Invention

[0006] The purpose of this invention is to provide a data-driven high-precision control method for micro-vibrations of synchrotron radiation sources, in order to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a data-driven high-precision control method for micro-vibrations of a synchrotron radiation source, comprising the following steps: S1. Data Acquisition and Preprocessing: Collect various data during the operation of the synchrotron radiation source, including but not limited to vibration data, beam energy, and experimental modes, and perform preprocessing operations such as filtering and denoising on the acquired raw data; Preprocessing process: First, outlier removal is performed using the 3σ criterion or an outlier detection algorithm based on density clustering; Second, data standardization is implemented, mapping parameters of different dimensions to the [0,1] interval; Finally, time windows are divided, and time-series samples are generated at 10-second intervals to ensure that each sample contains a complete parameter sequence; S2. Operating Condition Identification and Classification: The collected operating parameters are input into the operating condition identification model in real time. The model is used to identify and classify the current operating conditions of the light source, such as beam energy adjustment and experimental mode switching. S3. Control strategy parameter adjustment: Based on the identified working condition type, automatically adjust the parameters of the control algorithm (such as control frequency, control force, etc.) to adapt to the control requirements under different working conditions. S4. High-precision micro-vibration control: The PID control algorithm, combined with real-time parameter adjustment, is used to accurately compensate for the micro-vibrations of the synchrotron radiation source, ensuring the stability and accuracy of the source. S5. Performance Evaluation and Feedback Optimization: The control effect is evaluated in real time, and the control strategy is fine-tuned or the operating condition recognition model is retrained based on the evaluation results to improve control accuracy and stability.

[0008] Furthermore, in step S1, rigid installation (such as magnetic base or threaded fixation) is used to ensure the transmission of high-frequency vibration. High-precision vibration sensors are installed in the vibration-sensitive areas of the synchrotron radiation source (such as the storage ring vacuum chamber and the front end of the beamline). The sensors are used to monitor the vibration of the synchrotron radiation source in real time, and at the same time, the operating parameters such as beam energy and experimental mode are obtained through the control system.

[0009] Furthermore, the vibration-sensitive area includes: storage ring, insert, beamline, experimental station, supporting infrastructure, and external vibration source. The sensor installation locations in each area are as follows: 1) Storage ring: Magnet system: magnet support base (monitors overall magnet vibration, such as quadrupole magnets, hexapole magnets, and bending magnets), cooling water pipe interface (monitors vibration caused by thermal stress or water flow pulsation), magnet fixing bolts (detects loose mechanical connections or resonance), sensor type (high-frequency accelerometer, such as PCB 356A01, frequency response 0.5Hz-10kHz); Beam Position Monitor (BPM): Connection between BPM support and vacuum tube, BPM electronics enclosure (if coupled to mechanical structure); 2) Inserts: Oscillator / Gyroscope: Magnetic pole gap adjustment mechanism (monitors the relative displacement of magnetic pole pairs, requiring ±0.1μm), support frame (detects overall structural vibration); 3) Beamline: Front end area: collimator / blinder guide rail (monitors vibration of mechanical drive system), primary slit (detects high-frequency vibration of slit blade edge); Monochromator: crystal rotation axis (monitoring angular vibration), cooling plate contact surface (detecting vibration caused by thermal deformation, requiring temperature compensation), sensor type (laser interferometer or fiber optic displacement sensor); The reflector system includes: mirror support points (sensors deployed at 200mm intervals) and piezoelectric actuator (PZT) base (for monitoring feedback noise from the active adjustment system). 4) Experimental Station: Sample stage: Nano-positioning stage base (for direct monitoring of vibration in the sample area, requiring <1nm RMS), vibration-damping air flotation platform (for detecting leakage or failure of the vibration isolation system). Detectors: CCD detector bracket (to monitor image blurring caused by mechanical vibration), liquid nitrogen cooling pipe interface (to detect vibration transmission from cryogenic pump); 5) Supporting infrastructure: Foundation and seismic isolation system: Seismic isolation platform anchor points (to compare the vibration transmissibility above and below the platform), building structural columns (to monitor the ground vibration transmission path, with points spaced 10m apart); Auxiliary equipment: vacuum pump base (for vibrational spectrum analysis of turbomolecular pumps), cooling water pipe support (for detecting water flow pulsation); 6) External vibration source: The exterior ground of the building (monitoring traffic vibrations) and the base of the air conditioning unit (detecting unbalanced vibrations of the fan impeller).

[0010] Furthermore, the vibration sensor includes an accelerometer and a displacement sensor, as detailed below: Accelerometer: Measures the acceleration changes of key parts of the synchrotron radiation source, calculates vibration velocity and displacement through integration, and reflects the dynamic characteristics of micro-vibration. In actual operation, it is installed at key nodes of the light source structure to collect acceleration signals in real time and transmit them to the data acquisition system. Displacement sensor: directly measures minute displacement changes of the light source component, quantifies vibration amplitude and direction, and provides accurate compensation basis for control algorithm. In actual operation, laser displacement sensor or inductive displacement sensor is used to monitor the relative displacement of key components in a non-contact manner.

[0011] Furthermore, in step S2, the working condition identification model is trained using historical data and can output the corresponding working condition type based on the input operating parameters. The construction process of the working condition identification model is as follows: Historical data acquisition: Collect historical operating data of the synchrotron radiation source, including beam flux, electron beam trajectory offset, vacuum level, cooling system temperature and vibration acceleration, and perform outlier removal, standardization and time-series alignment on the raw data; Feature extraction: The sliding window method is used to extract time-domain features (mean, variance, peak factor) and frequency-domain features (the main frequency components are calculated by fast Fourier transform). Principal component analysis (PCA) is applied to reduce dimensionality, retaining the principal components with a cumulative contribution rate of over 95%, and compressing the original 20-dimensional parameters into a 5-7 dimensional feature space. Model architecture design and training: LSTM was selected as the basic model, a network architecture suitable for time series classification was designed, and the model was trained using labeled historical data. Cross-entropy loss and Adam optimizer were used for parameter optimization. Model validation and dynamic optimization: First, 10-fold cross-validation was performed to ensure the stability of the model on different data subsets; second, real-world simulation tests were conducted, and the model was deployed on the HEPS device for 72 hours of continuous operation to verify its response speed to sudden vibration events (<50ms); finally, an online learning mechanism was implemented, continuously incorporating new operating data through a sliding window update strategy, and automatically fine-tuning the model parameters every 24 hours. Deployment and integration: The trained model is packaged into a Docker container and deployed to the light source control system through a Kubernetes cluster to achieve real-time operating condition recognition and classification.

[0012] Furthermore, the operating condition types include the following types: Steady-state operating conditions: The electron beam current is stable (e.g., 100mA±0.1%), the magnet power supply fluctuation is <0.01%, and the main frequency of the vibration spectrum is concentrated in 1-10Hz (dominated by micro-vibration of the foundation), which is an ideal working condition; Beam tuning / injection conditions: During electron beam injection or track correction, the magnet current undergoes a step change (ΔI / I ~ 10⁻³), triggering transient mechanical vibrations of 50-200Hz, with vibration amplitudes reaching 5-10 times that of the steady state. Equipment start-up and shutdown conditions: Vacuum pumps / cooling units generate 20-500Hz wideband vibrations during start-up and shutdown (characteristic peaks appear at the blade passing frequency), and the vibration energy is 15-20dB higher than that in steady state. Abnormal vibration conditions: including special vibration modes caused by equipment failures such as magnet cooling water pipe resonance (characteristic frequency 80-120Hz) and insertion component transmission mechanism step loss (5-15Hz low frequency jitter); Environmental interference conditions: Enhanced foundation vibration caused by external vibration sources (such as nearby construction or traffic) (vibration amplitude in the 0.5-30Hz frequency band exceeds the standard by 2-5 times), with obvious time correlation; Mixed disturbance conditions: Multiple vibration sources coupled state (such as the start-up and shutdown of injection superimposed equipment), the vibration spectrum exhibits multi-peak characteristics, and the energy distribution of each frequency band changes dynamically with time.

[0013] Furthermore, in step S3, the optimal combination of control parameters under different operating conditions is determined in advance through experiments or simulations, and a control strategy parameter library is established. When the operating conditions change, the corresponding parameter combination is queried from the parameter library and selected for adjustment according to the current operating condition type. That is, the selected parameter combination is applied to the control algorithm to realize the dynamic adjustment of the control strategy.

[0014] Furthermore, step S4 includes the following sub-steps: S41. Input the adjusted control strategy parameters into the PID control algorithm; S42. The control algorithm calculates the required control quantities, such as compensation force and compensation displacement, based on the real-time collected vibration data and the preset control target. S43. The actuator (such as a motor, piezoelectric ceramic, etc.) converts the control quantity into actual physical action to compensate for the micro-vibration of the light source.

[0015] Furthermore, in step S5, a performance evaluation index system is established, and the control effect is quantified by comparing vibration data and experimental data before and after control. At the same time, the evaluation results are transmitted back to the control strategy parameters for adjustment using a feedback mechanism, so as to achieve continuous optimization of the control strategy.

[0016] This invention provides a data-driven high-precision control method for micro-vibrations of a synchrotron radiation source, which has the following advantages: This invention, driven by data, can identify and classify the operating conditions of light sources in real time. When the operating conditions of the light source change, it can automatically adjust the parameters of the control strategy without manual recalibration, achieving adaptive high-precision control of micro-vibrations under different operating conditions. This avoids the problem of decreased control accuracy that occurs with traditional fixed control strategies when operating conditions change. Furthermore, since the optimal combination of control parameters for different operating conditions is predetermined, when the operating conditions change, the corresponding parameter combination can be quickly selected from the parameter library for adjustment. This significantly reduces control adjustment time. Combined with a real-time performance evaluation and feedback optimization mechanism, this further ensures the stability and accuracy of the control effect. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the steps of a data-driven high-precision control method for micro-vibrations of a synchrotron radiation source according to the present invention. Detailed Implementation

[0018] The embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and should not be construed as limiting the scope of the invention.

[0019] like Figure 1 As shown, a data-driven high-precision control method for micro-vibrations of a synchrotron radiation source includes the following steps: S1. Data Acquisition and Preprocessing: Collect various data during the operation of the synchrotron radiation source, including but not limited to vibration data, beam energy, and experimental modes, and perform preprocessing operations such as filtering and denoising on the acquired raw data to provide a basis for subsequent analysis.

[0020] In this step, rigid installation (such as magnetic base or threaded fixation) is used to ensure the transmission of high-frequency vibrations. High-precision vibration sensors are installed in vibration-sensitive areas of the synchrotron radiation source (such as the storage ring vacuum chamber and the beamline front end). These sensors monitor the vibration of the synchrotron radiation source in real time, while the control system acquires operating parameters such as beam energy and experimental mode. The vibration-sensitive areas include: the storage ring, insert, beamline, experimental station, supporting infrastructure, and external vibration sources. The sensor installation locations in each area are as follows: 1) Storage ring: Magnet system: magnet support base (monitors overall magnet vibration, such as quadrupole magnets, hexapole magnets, and bending magnets), cooling water pipe interface (monitors vibration caused by thermal stress or water flow pulsation), magnet fixing bolts (detects loose mechanical connections or resonance), sensor type (high-frequency accelerometer, such as PCB 356A01, frequency response 0.5Hz-10kHz).

[0021] Beam position monitor (BPM): Connection between BPM support and vacuum tube, BPM electronics enclosure (if coupled to mechanical structure).

[0022] 2) Inserts: Oscillator / torsion oscillator: magnetic pole gap adjustment mechanism (monitors the relative displacement of magnetic pole pairs, requiring ±0.1μm), support frame (detects overall structural vibration).

[0023] 3) Beamline: Front end area: collimator / blinder guide rail (monitors vibration of mechanical drive system), primary slit (detects high-frequency vibration of slit blade edge).

[0024] Monochromator: crystal rotation axis (monitoring angular vibration), cooling plate contact surface (detecting vibration caused by thermal deformation, requiring temperature compensation), sensor type (laser interferometer or fiber optic displacement sensor).

[0025] Reflector system: mirror support points (sensors deployed at 200mm intervals), piezoelectric actuator (PZT) base (to monitor feedback noise of the active adjustment system).

[0026] 4) Experimental Station: Sample stage: Nano-positioning stage base (for direct monitoring of vibration in the sample area, requiring <1nm RMS), vibration-damping air flotation platform (for detecting leakage or failure of the vibration isolation system).

[0027] Detectors: CCD detector bracket (to monitor image blurring caused by mechanical vibration), liquid nitrogen cooling pipe interface (to detect vibration transmission from cryogenic pump).

[0028] 5) Supporting infrastructure: Foundation and seismic isolation system: Seismic isolation platform anchor points (to compare the vibration transmission rate above and below the platform), building structural columns (to monitor the ground vibration transmission path, with points spaced 10m apart).

[0029] Auxiliary equipment: vacuum pump base (for vibration spectrum analysis of turbomolecular pumps), cooling water pipe support (for detecting water flow pulsation).

[0030] 6) External vibration source: The exterior ground of the building (monitoring traffic vibrations) and the base of the air conditioning unit (detecting unbalanced vibrations of the fan impeller).

[0031] In this embodiment, the vibration sensor includes an accelerometer and a displacement sensor. The accelerometer measures the acceleration changes of key components of the synchrotron radiation source, calculates vibration velocity and displacement through integration, and reflects the dynamic characteristics of micro-vibrations. In actual operation, it is installed at key nodes of the light source structure (such as support beams, optical element bases, etc.) to collect acceleration signals in real time and transmit them to the data acquisition system. The displacement sensor directly measures minute displacement changes of the light source components, quantifies vibration amplitude and direction, and provides accurate compensation basis for the control algorithm. In actual operation, a laser displacement sensor or an inductive displacement sensor is used to monitor the relative displacement of key components non-contactly.

[0032] In this embodiment, the preprocessing flow of the raw data is as follows: First, outlier removal is performed using the 3σ criterion or an outlier detection algorithm based on density clustering; second, data standardization is implemented, mapping parameters of different dimensions to the [0,1] interval; finally, time windows are divided, and time-series samples are generated at 10-second intervals to ensure that each sample contains a complete parameter sequence.

[0033] S2. Operating Condition Identification and Classification: The collected operating parameters are input into the operating condition identification model in real time, and the model is used to identify and classify the current operating condition of the light source.

[0034] In this step, the operating condition identification model is trained using historical data and can output the corresponding operating condition type based on the input operating parameters. The construction process of the operating condition identification model is as follows: Historical data acquisition: Collect historical operating data of the synchrotron radiation source, including beam flux, electron beam trajectory offset, vacuum level, cooling system temperature and vibration acceleration, and perform outlier removal, standardization and time-series alignment on the raw data.

[0035] Feature extraction: The sliding window method is used to extract time-domain features (mean, variance, peak factor) and frequency-domain features (the main frequency components are calculated by fast Fourier transform). Principal component analysis (PCA) is applied to reduce dimensionality, retaining the principal components with a cumulative contribution rate of over 95%, and compressing the original 20-dimensional parameters into a 5-7 dimensional feature space.

[0036] Model architecture design and training: LSTM was selected as the basic model, and a network architecture suitable for time series classification was designed. The model was trained using labeled historical data, and the parameters were optimized using cross-entropy loss and Adam optimizer.

[0037] Model validation and dynamic optimization: First, 10-fold cross-validation was performed to ensure the stability of the model on different subsets of data; second, real-world simulation tests were conducted, and the model was deployed on the HEPS device for 72 hours of continuous operation to verify its response speed to sudden vibration events (<50ms); finally, an online learning mechanism was implemented, continuously incorporating new operating data through a sliding window update strategy, and automatically fine-tuning the model parameters every 24 hours.

[0038] Deployment and integration: The trained model is packaged into a Docker container and deployed to the light source control system through a Kubernetes cluster to achieve real-time operating condition recognition and classification.

[0039] In this embodiment, the operating condition types include the following: Steady-state operating conditions: The electron beam current is stable (e.g., 100mA±0.1%), the magnet power supply fluctuation is <0.01%, and the main frequency of the vibration spectrum is concentrated in 1-10Hz (dominated by micro-vibration of the foundation), which is an ideal working condition; Beam tuning / injection conditions: During electron beam injection or track correction, the magnet current undergoes a step change (ΔI / I ~ 10⁻³), triggering transient mechanical vibrations of 50-200Hz, with vibration amplitudes reaching 5-10 times that of the steady state. Equipment start-up and shutdown conditions: Vacuum pumps / cooling units generate 20-500Hz wideband vibrations during start-up and shutdown (characteristic peaks appear at the blade passing frequency), and the vibration energy is 15-20dB higher than that in steady state. Abnormal vibration conditions: including special vibration modes caused by equipment failures such as magnet cooling water pipe resonance (characteristic frequency 80-120Hz) and insertion component transmission mechanism step loss (5-15Hz low frequency jitter); Environmental interference conditions: Enhanced foundation vibration caused by external vibration sources (such as nearby construction or traffic) (vibration amplitude in the 0.5-30Hz frequency band exceeds the standard by 2-5 times), with obvious time correlation; Mixed disturbance conditions: Multiple vibration sources coupled state (such as the start-up and shutdown of injection superimposed equipment), the vibration spectrum exhibits multi-peak characteristics, and the energy distribution of each frequency band changes dynamically with time.

[0040] S3. Control strategy parameter adjustment: Based on the identified operating condition type, automatically adjust the parameters of the control algorithm (such as control frequency, control force, etc.) to adapt to the control requirements under different operating conditions.

[0041] In this step, the optimal combination of control parameters under different operating conditions is determined in advance through experiments or simulations, and a control strategy parameter library is established. When the operating conditions change, the corresponding parameter combination is queried from the parameter library and selected for adjustment according to the current operating condition type. That is, the selected parameter combination is applied to the control algorithm to realize the dynamic adjustment of the control strategy.

[0042] S4. High-precision micro-vibration control: The PID control algorithm, combined with real-time parameter adjustment, is used to accurately compensate for the micro-vibrations of the synchrotron radiation source, ensuring the stability and accuracy of the source.

[0043] S41. Input the adjusted control strategy parameters into the PID control algorithm.

[0044] S42. The control algorithm calculates the required control quantities, such as compensation force and compensation displacement, based on the real-time collected vibration data and the preset control target.

[0045] S43. The actuator (such as a motor, piezoelectric ceramic, etc.) converts the control quantity into actual physical action to compensate for the micro-vibration of the light source.

[0046] S5. Performance Evaluation and Feedback Optimization: The control effect is evaluated in real time, and the control strategy is fine-tuned or the operating condition recognition model is retrained based on the evaluation results to improve control accuracy and stability.

[0047] In this step, a performance evaluation index system is established. By comparing vibration data and experimental data before and after control, the control effect is quantified. Simultaneously, a feedback mechanism is used to transmit the evaluation results back to the control strategy parameters for adjustment, achieving continuous optimization of the control strategy. The specific operation of this embodiment is as follows: First, performance evaluation indicators are set, such as the vibration amplitude reduction rate and the improvement rate of experimental data accuracy. Then, the values ​​of the performance evaluation indicators are calculated and displayed in real time to understand the current control effect. Finally, based on the performance evaluation results, the control strategy parameters are fine-tuned or the operating condition recognition model is retrained to further improve control accuracy and stability.

[0048] The embodiments of the present invention are given for illustrative and descriptive purposes only, and are not intended to be exhaustive or to limit the invention to the forms disclosed. Many modifications and variations will be apparent to those skilled in the art. The embodiments were chosen and described in order to better illustrate the principles and practical application of the invention, and to enable those skilled in the art to understand the invention and to design various embodiments with various modifications suitable for a particular purpose.

Claims

1. A data-driven high-precision control method for micro-vibrations of a synchrotron radiation source, characterized in that, Includes the following steps: S1. Data Acquisition and Preprocessing: Collect various data during the operation of the synchrotron radiation source, including but not limited to vibration data, beam energy, and experimental modes, and filter and denoise the acquired raw data; S2. Operating Condition Identification and Classification: The collected operating parameters are input into the operating condition identification model in real time, and the model is used to identify and classify the current operating condition of the light source. S3. Control strategy parameter adjustment: Automatically adjust the parameters of the control algorithm according to the identified working condition type; S4. High-precision micro-vibration control: The PID control algorithm, combined with real-time adjusted parameters, is used to accurately compensate for the micro-vibrations of the synchrotron radiation source.

2. The high-precision control method for micro-vibration of a synchrotron radiation source based on data-driven approach according to claim 1, characterized in that, In step S1, a vibration sensor is rigidly installed in the vibration-sensitive area of ​​the synchrotron radiation source. The sensor is used to monitor the vibration of the synchrotron radiation source in real time, and the control system obtains operating parameters such as beam energy and experimental mode.

3. The high-precision control method for micro-vibration of a synchrotron radiation source based on data-driven approach according to claim 2, characterized in that, The vibration-sensitive area includes: storage ring, insert, beamline, experimental station, supporting infrastructure, and external vibration source. The sensor installation locations in each area are as follows: 1) Storage ring: Magnet system: magnet support base, cooling water pipe interface, magnet fixing bolts, sensor type; Beam position monitor: connection between BPM support and vacuum tube, BPM electronics chassis; 2) Inserts: Oscillator / Gyroscope: Magnetic pole gap adjustment mechanism and support frame; 3) Beamline: Front end area: collimator / blocker guide rail, primary slit; Monochromator: crystal rotation axis, cooling plate contact surface, sensor type; Reflector system: mirror support point, piezoelectric actuator base; 4) Experimental station: Sample stage: Nano-positioning stage base, vibration-damping air flotation platform; Detector: CCD detector bracket, liquid nitrogen cooling pipe interface; 5) Supporting infrastructure: Foundation and seismic isolation system: seismic isolation platform anchor points, building structural columns; Auxiliary equipment: vacuum pump base, cooling water pipe support; 6) External vibration source: The exterior ground of the building and the base of the air conditioning unit.

4. The high-precision control method for micro-vibration of a synchrotron radiation source based on data-driven approach according to claim 2, characterized in that, The vibration sensor includes an accelerometer and a displacement sensor, as detailed below: Accelerometer: Measures the acceleration changes of key parts of a synchrotron radiation source, and calculates the vibration velocity and displacement through integration, reflecting the dynamic characteristics of micro-vibration; Displacement sensor: directly measures minute displacement changes of the light source component, quantifying the vibration amplitude and direction.

5. The high-precision control method for micro-vibration of a synchrotron radiation source based on data-driven approach according to claim 1, characterized in that, In step S2, the working condition identification model is trained using historical data and can output the corresponding working condition type based on the input operating parameters. The construction process of the working condition identification model is as follows: Historical data acquisition: Collect historical operating data of the synchrotron radiation source, including beam flux, electron beam trajectory offset, vacuum level, cooling system temperature and vibration acceleration, and perform outlier removal, standardization and time-series alignment on the raw data; Feature extraction: The sliding window method is used to extract time-domain and frequency-domain features, and principal component analysis is applied for dimensionality reduction. Model architecture design and training: LSTM was selected as the basic model, a network architecture suitable for time series classification was designed, and the model was trained using labeled historical data. Cross-entropy loss and Adam optimizer were used for parameter optimization. Model validation and dynamic optimization: First, 10-fold cross-validation was performed; second, real-world operating condition simulation tests were conducted, and the model was deployed on the HEPS device for 72 hours of continuous operation to verify its response speed to sudden vibration events; finally, an online learning mechanism was implemented, continuously incorporating new operating data through a sliding window update strategy, and automatically fine-tuning the model parameters every 24 hours. Deployment and integration: The trained model is packaged into a Docker container and deployed to the light source control system through a Kubernetes cluster to achieve real-time operating condition recognition and classification.

6. The high-precision control method for micro-vibration of a synchrotron radiation source based on data-driven technology according to claim 5, characterized in that, The operating condition types include the following: Steady-state operating conditions: The electron beam current is stable, the magnet power supply fluctuation is <0.01%, and the main frequency of the vibration spectrum is concentrated in 1-10Hz; Beam modulation / injection conditions: During electron beam injection or orbit correction, the magnet current changes abruptly, causing transient mechanical vibrations of 50-200Hz, with vibration amplitudes reaching 5-10 times that of the steady state. Equipment start-up and shutdown conditions: Vacuum pumps / cooling units generate 20-500Hz wideband vibrations when starting and stopping, with vibration energy 15-20dB higher than that in steady state; Abnormal vibration conditions: including special vibration modes caused by equipment failures such as magnet cooling water pipe resonance and loss of step in the insert transmission mechanism; Environmental disturbance conditions: Enhanced foundation vibration caused by external vibration sources, with significant time correlation; Mixed disturbance condition: Multiple vibration sources are coupled, the vibration spectrum exhibits multi-peak characteristics, and the energy distribution of each frequency band changes dynamically with time.

7. The high-precision control method for micro-vibration of a synchrotron radiation source based on data-driven approach according to claim 1, characterized in that, In step S3, the optimal combination of control parameters under different operating conditions is determined in advance through experiments or simulations, and a control strategy parameter library is established. When the operating conditions change, the corresponding parameter combination is queried from the parameter library and selected for adjustment according to the current operating condition type. That is, the selected parameter combination is applied to the control algorithm to realize the dynamic adjustment of the control strategy.

8. The high-precision control method for micro-vibration of a synchrotron radiation source based on data-driven approach according to claim 1, characterized in that, Step S4 includes the following sub-steps: S41. Input the adjusted control strategy parameters into the PID control algorithm; S42. The control algorithm calculates the required control quantity based on the real-time collected vibration data and the preset control target. S43. The actuator converts the control quantity into actual physical action to compensate for the micro-vibration of the light source.

9. The high-precision control method for micro-vibration of a synchrotron radiation source based on data-driven approach according to claim 1, characterized in that, This includes the following steps: S5. Performance Evaluation and Feedback Optimization: The control effect is evaluated in real time, and the control strategy is fine-tuned or the operating condition recognition model is retrained based on the evaluation results.

10. A high-precision control method for micro-vibration of a synchrotron radiation source based on data-driven approach according to claim 9, characterized in that, In step S5, a performance evaluation index system is established. By comparing vibration data and experimental data before and after control, the control effect is quantified. At the same time, the evaluation results are transmitted back to the control strategy parameters for adjustment using a feedback mechanism, so as to achieve continuous optimization of the control strategy.

Images