Laser ion source system based on double pulses and AI feedback control and control method

Abstract

The invention discloses a laser ion source system based on double pulses and AI feedback control and a control method, and relates to the technical field of plasmas, and the system is characterized in that a first laser generates pre-ablation laser, and generates pre-plasma on a target material; the second laser generates main ionization laser which irradiates the pre-plasma to generate main plasma; and the intelligent control unit extracts spatial features from the plume image, fuses the spatial features, the spectral data, the real-time ion current intensity and the equipment parameters to obtain high-order state features, and generates laser pulse optimization parameters according to the high-order state features and the equipment parameters. According to the invention, a double-pulse mechanism is adopted to enable main ionization to occur in the expanded plasma, charge exchange and energy dispersion are reduced, the energy divergence of the ion beam is remarkably reduced, the energy conversion efficiency is improved, AI real-time feedback control can compensate factors such as target pitting corrosion and laser power fluctuation, and very high inter-pulse stability is realized.

Description

Technical Field

[0001] This application relates to the field of plasma technology, and in particular to a system for generating plasma using dual-pulse laser and AI (artificial intelligence) feedback control technology. Background Technology

[0002] Laser ion sources are key front-end devices in high-performance accelerators. Traditional single-pulse laser ion sources produce plasmas with large energy dissipation and wide pulse broadening, and suffer from shot-to-shot instability, leading to low transport and acceleration efficiency in subsequent accelerators. Although magnetron sputtering and arc discharge plasma source technologies are relatively mature, they are difficult to generate ions of certain special elements or extremely high charge states. Therefore, developing a laser ion source that can simultaneously achieve high ionization efficiency, low energy dissipation, and excellent stability is of great significance. Summary of the Invention

[0003] This application provides a laser ion source system and control method based on dual-pulse and AI feedback control to solve the problem that existing laser ion sources are unable to simultaneously achieve high ionization efficiency, low energy dissipation, and excellent stability.

[0004] On one hand, embodiments of this application provide a laser ion source system based on dual-pulse and AI feedback control, including: The first laser is used to generate the pre-ablation laser; The second laser is used to generate the main ionization laser after a preset delay following the pre-ablation laser generated by the first laser. When the pre-ablation laser irradiates the target material, a pre-plasma is generated. After the main ionization laser irradiates the pre-plasma, the main plasma is generated. High-speed cameras and spectrometers are used to acquire plume images and spectral data of the pre-plasma and main plasma; The beam diagnostic unit is used to acquire the real-time ion current intensity during the main plasma transport process; The parameter acquisition unit is used to acquire equipment parameters, including real-time energy data and pulse width data of the first laser and the second laser, as well as the position data of the target material. The intelligent control unit is used to extract spatial features from the feather image using a diagnostic inference model. The diagnostic inference model also fuses the spatial features, spectral data, real-time ion current intensity, and equipment parameters to obtain higher-order state features. The intelligent control unit also uses a parameter optimization model to generate laser pulse optimization parameters based on the higher-order state features and equipment parameters. The laser pulse optimization parameters include the delay of the main ionizing laser relative to the pre-ablation laser, the energy of the main ionizing laser, and the focal position of the pre-ablation laser.

[0005] On the other hand, embodiments of this application also provide a laser ion source control method based on dual-pulse and AI feedback control, including: A pre-ablation laser is generated, which irradiates the target material to produce pre-plasma. After generating the pre-ablation laser, a preset delay is applied to generate the main ionization laser, which then irradiates the pre-plasma to generate the main plasma. Acquire plume images and spectral data of the pre-plasma and main plasma; Real-time ion flow intensity during the main plasma transport process was collected; Acquire equipment parameters, including real-time energy data and pulse width data of the pre-ablation laser and the main ionization laser, as well as the position data of the target material; A diagnostic reasoning model is used to extract spatial features from feather images. Spatial features, spectral data, real-time ion current intensity, and equipment parameters are fused to obtain higher-order state features. A parameter optimization model is used to generate laser pulse optimization parameters based on higher-order state characteristics and equipment parameters. The laser pulse optimization parameters include the delay of the main ionizing laser relative to the pre-ablation laser, the energy of the main ionizing laser, and the focal position of the pre-ablation laser.

[0006] The laser ion source system and control method based on dual-pulse and AI feedback control in this application have the following advantages: 1. Low energy dispersion. The dual-pulse mechanism causes the main ionization to occur in the expanded rarefied plasma, reducing charge exchange and energy dispersion, and significantly reducing the energy dispersion of the ion beam.

[0007] 2. High stability. AI real-time feedback control can compensate for factors such as target erosion and laser power fluctuations, achieving high inter-pulse stability.

[0008] 3. High efficiency. Pre-plasma provides better absorption conditions for the main laser, improving the conversion efficiency of laser energy to ion beam. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a schematic diagram of the composition of a laser ion source system based on dual-pulse and AI feedback control provided in an embodiment of this application.

[0011] Figure 2This is an architecture diagram of a laser ion source control method based on dual-pulse and AI feedback control provided in an embodiment of this application.

[0012] Explanation of reference numerals: 1. Vacuum chamber; 2. First laser; 3. Pre-ablation laser; 4. Target material; 5. Pre-plasma; 6. Second laser; 7. Main ionization laser; 8. Main plasma; 9. High-speed camera and spectrometer; 10. Intelligent control unit; 11. Faraday cylinder; 12. Electrostatic analyzer. Detailed Implementation

[0013] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0014] Figure 1 This is a schematic diagram illustrating the composition of a laser ion source system based on dual-pulse and AI feedback control, as provided in an embodiment of this application. The embodiment of this application provides a laser ion source system based on dual-pulse and AI feedback control, comprising: First laser 2, used to generate pre-ablation laser 3; The second laser 6 is used to generate the main ionization laser 7 after a preset delay following the generation of the pre-ablation laser 3 by the first laser 2. When the pre-ablation laser 3 irradiates the target material 4, a pre-plasma 5 is generated. After the main ionization laser 7 irradiates the pre-plasma 5, a main plasma 8 is generated. High-speed camera and spectrometer 9 are used to acquire plume images and spectral data of pre-plasma 5 and main plasma 8; The beam intensity and time structure were measured using a Faraday tube 11; the charge state was analyzed using the time-of-flight method with an electrostatic analyzer 12, and the charge state distribution of the beam was measured.

[0015] The parameter acquisition unit is used to acquire equipment parameters, including real-time energy data and pulse width data of the first laser 2 and the second laser 6, as well as the position data of the target 4. The intelligent control unit 10 is used to extract spatial features from the feather image using a diagnostic reasoning model. The diagnostic reasoning model also fuses the spatial features, spectral data, real-time ion current intensity, and equipment parameters to obtain higher-order state features. The intelligent control unit 10 also uses a parameter optimization model to generate laser pulse optimization parameters based on the higher-order state features and equipment parameters. The laser pulse optimization parameters include the delay of the main ionizing laser 7 relative to the pre-ablation laser 3, the energy of the main ionizing laser 7, and the focal position of the pre-ablation laser 3.

[0016] For example, the first laser 2 and the second laser 6 are pulsed lasers that are time-synchronized and whose wavelengths can be independently adjusted. The first laser 2 generates a pre-ablation laser 3 with a pulse length on the order of nanoseconds, and the second laser 6 generates a main ionization laser 7 with a pulse length on the order of picoseconds or femtoseconds. The pre-ablation laser 3 and the main ionization laser 7 will irradiate the same position on the target material 4 so that most of the energy of the main ionization laser 7 can be accurately applied to the sparse pre-plasma 5. Moreover, when the main ionization laser 7 generated later acts on the pre-plasma 5 that has already been formed, it can efficiently and gently ionize the pre-plasma 5 to generate a high-quality main plasma 8.

[0017] Furthermore, the delay of the first laser 2 relative to the second laser 6 can be controlled by the intelligent control unit 10 in the range of tens to hundreds of nanoseconds.

[0018] The beam diagnostic unit is also used to collect beam parameters during the transmission of the main plasma 8. A Faraday cup is used to collect real-time ion current intensity and ion beam pulse time structure.

[0019] The first laser 2, the second laser 6, and the target material 4 are all located in the vacuum chamber 1.

[0020] In the embodiments of this application, the intelligent control unit 10 completes the process of processing all data based on an AI decision engine, and the data processing process is as follows: Figure 2 As shown, the AI ​​decision engine includes a multimodal data perception layer and a plasma dynamic digital twin. The plasma dynamic digital twin includes a diagnostic inference model and a parameter optimization model. The multimodal data perception layer acquires multimodal data required for subsequent processing from the high-speed camera and spectrometer 9, the beam diagnostic unit, and the parameter acquisition unit. This multimodal data includes plume images acquired by the high-speed camera, spectral data acquired by the spectrometer, real-time ion current intensity acquired by the beam diagnostic unit, and equipment parameters acquired by the parameter acquisition unit. These plume images are used to extract spatial features of the plasma (including pre-plasma 5 and main plasma 8), such as expansion size, symmetry, and luminescence intensity distribution. The spectral data is used to identify ion species and charge distribution. After obtaining the spectral data, the electron density Ne and electron temperature Te can be retrieved based on the Stark broadening and spectral line intensity ratio.

[0021] The diagnostic reasoning model includes a convolutional neural network (CNN) and a Transformer module. The CNN is used to extract spatial features from the feather image, and the Transformer module is used to fuse spatial features, spectral data, and device parameters to obtain higher-order state features.

[0022] The diagnostic reasoning model in this embodiment adopts a hybrid architecture of CNN and Transformer, which can realize multimodal fusion diagnosis of the current plasma state and obtain the required high-order state features. These high-order state features may include plasma density gradient, predicted coupling efficiency of main ionizing laser 7 and main plasma 8, and soft measurement value of expected ion energy dissipation.

[0023] The parameter optimization model uses deep reinforcement learning (DRL) technology to generate laser pulse optimization parameters.

[0024] Specifically, in the deep reinforcement learning framework, the state space consists of the multimodal data received by the multimodal data perception layer, and the action space consists of the laser pulse optimization parameters for the next time step. Furthermore, the reward function used in the parameter optimization model is expressed as: R = w 1· I ic - w 2· Ed - w 3· A pa in, R For real-time rewards, I ic For real-time ion current intensity. Ed For ion energy divergence, A pa For parameter adjustment range, w 1. w 2 and w 3 represents the weighting coefficient for each item. As can be seen from the above reward function, the parameter optimization model in this embodiment of the application can not only increase the ion current intensity but also actively suppress energy dissipation and avoid drastic parameter oscillations, achieving a smooth optimization effect.

[0025] As can be seen, in the embodiments of this application, the intelligent control unit uses the current state data of the laser and the target to calculate the action data at the next moment during the control process, thereby realizing closed-loop feedback control.

[0026] In the embodiments of this application, the AI ​​decision engine (mainly a plasma dynamic digital twin) can accumulate massive training data in a laboratory environment through tens of thousands of "targeting-measuring" cycles to train the diagnostic inference model and parameter optimization model offline, enabling these two models to learn the dynamic laws of plasma and the optimal control strategy.

[0027] Once a trained and effective AI decision engine is obtained, it can be lightweighted and then deployed on an edge computing device at the ion source equipment to meet the requirements of microsecond-level real-time inference latency.

[0028] The system in this embodiment has online fine-tuning capability. That is, when the system performance is detected to be slowly degrading due to equipment aging or target material replacement, a safe mode can be activated to perform small-batch online learning within a limited range, enabling the AI ​​decision engine to adapt to the slowly changing environment.

[0029] This application also provides a laser ion source control method based on dual-pulse and AI feedback control, including the following steps: A pre-ablation laser 3 is generated, and the pre-ablation laser 3 irradiates the target material 4 to generate pre-plasma 5. After the pre-ablation laser 3 is generated, the main ionization laser 7 is generated after a preset delay. The main ionization laser 7 irradiates the pre-plasma 5 to generate the main plasma 8. Feather images and spectral data of pre-plasma 5 and main plasma 8 were acquired; The real-time ion flow intensity during the transport process of the main plasma 8 was collected; Acquire equipment parameters, including real-time energy data and pulse width data of pre-ablation laser 3 and main ionization laser 7, as well as position data of target 4; A diagnostic reasoning model is used to extract spatial features from feather images. Spatial features, spectral data, real-time ion current intensity, and equipment parameters are fused to obtain higher-order state features. A parameter optimization model is used to generate laser pulse optimization parameters based on higher-order state characteristics and equipment parameters. The laser pulse optimization parameters include the delay of the main ionizing laser 7 relative to the pre-ablation laser 3, the energy of the main ionizing laser 7, and the focal position of the pre-ablation laser 3.

[0030] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0031] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A laser ion source system based on dual-pulse and AI feedback control, characterized in that, include: A first laser (2) is used to generate a pre-ablation laser (3); The second laser (6) is used to generate the main ionization laser (7) after a preset delay following the generation of the pre-ablation laser (3) by the first laser (2). When the pre-ablation laser (3) irradiates the target material (4), a pre-plasma (5) is generated, and the main ionization laser (7) irradiates the pre-plasma (5) to generate the main plasma (8). A high-speed camera and spectrometer (9) are used to acquire plume images and spectral data of the pre-plasma (5) and the main plasma (8); A beam diagnostic unit is used to collect the real-time ion flow intensity during the transmission process of the main plasma (8); The parameter acquisition unit is used to acquire device parameters, which include real-time energy data and pulse width data of the first laser (2) and the second laser (6) and position data of the target material (4); The intelligent control unit (10) is used to extract spatial features from the feather image using a diagnostic reasoning model. The diagnostic reasoning model also fuses the spatial features, the spectral data, the real-time ion current intensity, and the device parameters to obtain higher-order state features. The intelligent control unit (10) also uses a parameter optimization model to generate laser pulse optimization parameters based on the higher-order state features and the device parameters. The laser pulse optimization parameters include the delay of the main ionizing laser (7) relative to the pre-ablation laser (3), the energy of the main ionizing laser (7), and the focal position of the pre-ablation laser (3).

2. The laser ion source system based on dual-pulse and AI feedback control according to claim 1, characterized in that, The first laser (2) generates the pre-ablation laser (3) with a pulse length on the order of nanoseconds, and the second laser (6) generates the main ionization laser (7) with a pulse length on the order of picoseconds or femtoseconds.

3. The laser ion source system based on dual-pulse and AI feedback control according to claim 1, characterized in that, It also includes an ion extraction unit for extracting the main plasma (8).

4. The laser ion source system based on dual-pulse and AI feedback control according to claim 1, characterized in that, The beam diagnostic unit is also used to collect position information of the main plasma (8) during its transmission process.

5. The laser ion source system based on dual-pulse and AI feedback control according to claim 4, characterized in that, The beam diagnostic unit comprises a Faraday tube and a beam position detector. The Faraday tube is used to acquire the real-time ion current intensity, and the beam position detector is used to acquire the position information.

6. The laser ion source system based on dual-pulse and AI feedback control according to claim 1, characterized in that, The first laser (2), the second laser (6) and the target (4) are all disposed in the vacuum chamber (1).

7. The laser ion source system based on dual-pulse and AI feedback control according to claim 1, characterized in that, The diagnostic reasoning model includes a convolutional neural network and a Transformer module. The convolutional neural network is used to extract the spatial features from the feather image, and the Transformer module is used to fuse the spatial features, the spectral data, and the device parameters to obtain the higher-order state features.

8. The laser ion source system based on dual-pulse and AI feedback control according to claim 1, characterized in that, The parameter optimization model uses deep reinforcement learning technology to generate the laser pulse optimization parameters.

9. The laser ion source system based on dual-pulse and AI feedback control according to claim 8, characterized in that, The reward function used in the parameter optimization model is expressed as follows: R = w 1· I ic - w 2· Ed - w 3· A pa in, R For real-time rewards, I ic The real-time ion current intensity is [value missing]. Ed For ion energy divergence, A pa For parameter adjustment range, w 1. w 2 and w 3 represents the weighting coefficient for each item.

10. A control method applied to the laser ion source system based on dual-pulse and AI feedback control as described in any one of claims 1-9, characterized in that, include: A pre-ablation laser (3) is generated, which irradiates the target material (4) to generate pre-plasma (5). After the pre-ablation laser (3) is generated, a main ionization laser (7) is generated after a preset delay. The main ionization laser (7) irradiates the pre-plasma (5) to generate the main plasma (8). Collect plume images and spectral data of the pre-plasma (5) and the main plasma (8); The real-time ion flow intensity during the transmission process of the main plasma (8) is collected; Acquire equipment parameters, including real-time energy data and pulse width data of the pre-ablation laser (3) and the main ionization laser (7) and position data of the target material (4); A diagnostic reasoning model is used to extract spatial features from the plume image. The spatial features, the spectral data, the real-time ion current intensity, and the device parameters are fused to obtain higher-order state features. A parameter optimization model is used to generate laser pulse optimization parameters based on the higher-order state characteristics and the device parameters. The laser pulse optimization parameters include the delay of the main ionizing laser (7) relative to the pre-ablation laser (3), the energy of the main ionizing laser (7), and the focal position of the pre-ablation laser (3).

Images