Intelligent control system of neutral beam injection device

Abstract

The invention discloses an intelligent control system for a neutral beam injection device, and the system comprises a measurement and control module which is used for controlling a plurality of subsystems of the neutral beam injection device to work cooperatively in a neutral beam injection experiment process, and carrying out the management of experiment data generated in the neutral beam injection experiment process; the digital twinning module is used for constructing a digital twinning body of the neutral beam injection device in the digital space and mapping the operation states of a plurality of subsystems of the neutral beam injection device according to the digital twinning body; and the performance prediction module is used for predicting the possibility of sparking in the neutral beam injection experiment process and predicting the neutral beam injection power of the neutral beam injection device under the current experiment parameters. By adopting the technical scheme of the invention, the safety, reliability, timeliness and intelligent level of controlling the neutral beam injection device can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of large electrophysical equipment control, and particularly to an intelligent control system for a neutral beam injection device. Background Art

[0002] With the continuous growth of the world's population and the rapid development of the economy, the global demand for energy is also increasing continuously. Nuclear fusion technology has shown great development potential due to its abundant raw materials, high output efficiency, and higher safety compared to nuclear fission, and is regarded as one of the ideal energy forms for the future.

[0003] Tokamak is a toroidal magnetic field device that uses magnetic confinement to achieve controlled nuclear fusion. It generates a toroidal magnetic field to confine the plasma and encloses the high-temperature plasma in a toroidal container. In the nuclear fusion reaction using deuterium-tritium fuel, the Lawson criterion needs to be met to achieve the fusion reaction. The Lawson criterion generally requires the plasma temperature to reach 10 - 20 keV. However, the tokamak device usually uses ohmic heating, and the heating efficiency of the plasma decreases as the electron temperature increases, and can only make the plasma temperature reach about 3 keV, resulting in the plasma temperature not meeting the requirements of the deuterium-tritium fusion reaction in large-scale devices.

[0004] In response to the above problems, auxiliary heating technologies such as electron cyclotron, ion cyclotron, lower hybrid wave, and neutral beam injection have emerged. Neutral beam injection heating is the auxiliary heating method with the clearest physical mechanism and the most obvious heating effect in the tokamak heating means. Its main function is to inject high-energy neutral beams into the tokamak device, collide with the background plasma of the tokamak device for energy exchange, and then achieve the purpose of heating the background plasma of the tokamak device. In order to better achieve the heating purpose, the control of the neutral beam injection device is particularly important. However, the existing control schemes for neutral beam injection devices are relatively backward. Especially with the continuous in-depth development of neutral beam injection experiments, higher requirements are put forward in terms of the safety, reliability, timeliness, and intelligent level of the control schemes for neutral beam injection devices. Summary of the Invention

[0005] The purpose of the embodiments of the present invention is to provide an intelligent control system for a neutral beam injection device. By integrating the control and management of the experimental process of the neutral beam injection device, using a digital twin to map the entity operation state, and predicting the performance of the neutral beam injection device for arcing and key experimental parameters, the safety, reliability, timeliness, and intelligent level of controlling the neutral beam injection device can be improved.

[0006] To achieve the above object, an embodiment of the present invention provides an intelligent control system for a neutral beam injection device, including a measurement and control module, a digital twin module, and a performance prediction module; The measurement and control module is used to control the coordinated operation of multiple subsystems of the neutral beam injection device during the neutral beam injection experiment, and manage the experimental data generated during the neutral beam injection experiment; The digital twin module is used to construct a digital twin of the neutral beam injection device in the digital space, and map the operating states of multiple subsystems of the neutral beam injection device according to the digital twin; The performance prediction module is used to predict the possibility of arcing during the neutral beam injection experiment, and predict the neutral beam injection power of the neutral beam injection device under the current experimental parameters.

[0007] Further, the measurement and control module includes: An integrated control unit, which is used to adopt distributed control technology to control the coordinated operation of multiple subsystems of the neutral beam injection device during the neutral beam injection experiment, and is also used to centrally manage the experimental parameter configuration information and experimental control process of the neutral beam injection device; A data management unit, which is used to manage the experimental data generated during the neutral beam injection experiment; wherein, the management includes data acquisition, data storage, and data sharing.

[0008] Further, the data management unit manages the experimental data generated during the neutral beam injection experiment, specifically including: For the experimental data generated during the neutral beam injection experiment, low-speed acquisition and high-speed acquisition are respectively performed according to the data acquisition requirements; The experimental data acquired at low speed and the experimental data acquired at high speed are stored in a unified data storage format, and are stored in the short term during the experiment operation, and are converted to long-term storage after the experiment ends; wherein, the storage period of the short-term storage is less than the storage period of the long-term storage; Share the experimental data acquired at low speed and the experimental data acquired at high speed with the experimenters.

[0009] Further, the measurement and control module further includes: An interlock protection unit, which is used to perform interlock protection on the neutral beam injection device during the neutral beam injection experiment; wherein, the interlock protection uses optical signals as the transmission medium for fault signals.

[0010] Further, the digital twin module includes: A virtual-real data circulation unit, which is used to realize the two-way circulation of the actual operation data of multiple subsystems of the neutral beam injection device and the virtual simulation data of the digital twin. An actual data mapping unit for mapping the actual operation data of multiple subsystems of the neutral beam injection device into the digital space; An intelligent status recognition unit for recognizing the operation status of multiple subsystems of the neutral beam injection device and adjusting the operation status of the digital twin according to the recognition result.

[0011] Furthermore, the digital twin module further includes: A three-dimensional visualization unit for performing three-dimensional visualization processing on the operation status and data information of the digital twin.

[0012] Furthermore, the digital twin has a dual operation kernel, where one operation kernel is used to serve the mapping of the operation status of multiple subsystems of the neutral beam injection device, and the other operation kernel is used to serve the operation performance prediction of the neutral beam injection device.

[0013] Furthermore, the performance prediction module includes: A spark prediction unit for predicting the possibility of a spark occurring during a neutral beam injection experiment according to a trained spark prediction model; A power prediction unit for predicting the neutral beam injection power of the neutral beam injection device under current experimental parameters according to a trained power prediction model.

[0014] Furthermore, the performance prediction module further includes: A data processing unit for constructing a spark prediction data set and a power prediction data set to train a preset spark prediction model according to the spark prediction data set and train a preset power prediction model according to the power prediction data set.

[0015] Furthermore, the data processing unit constructs a spark prediction data set and a power prediction data set, specifically including: Selecting key parameters as data set features according to the operation principle of the neutral beam injection device; wherein, the key parameters include parameters of a radio frequency power source, an extraction power source, an acceleration power source, a bias power source, a magnetic field power source, and an intake air volume; Analyzing the historical experimental data corresponding to the data set features to obtain data set feature values; Filtering the data set feature values according to a preset filtering rule to construct a spark prediction data set and a power prediction data set.

[0016] Compared with the prior art, the embodiment of the present invention provides an intelligent control system for a neutral beam injection device, including a measurement and control module, a digital twin module, and a performance prediction module; wherein, the measurement and control module is used to control the coordinated operation of multiple subsystems of the neutral beam injection device during the neutral beam injection experiment and manage the experimental data generated during the neutral beam injection experiment; the digital twin module is used to construct a digital twin of the neutral beam injection device in the digital space and map the operating states of multiple subsystems of the neutral beam injection device according to the digital twin; the performance prediction module is used to predict the possibility of arcing during the neutral beam injection experiment and predict the neutral beam injection power of the neutral beam injection device under the current experimental parameters. The embodiment of the present invention can improve the safety and reliability of controlling the neutral beam injection device by integrating the control and management of the experimental process of the neutral beam injection device and predicting the performance of arcing and key experimental parameters of the neutral beam injection device. At the same time, the operating states of multiple subsystems of the neutral beam injection device are mapped by using the digital twin, which can improve the timeliness and intelligent level of controlling the neutral beam injection device, thus meeting the intelligent operation requirements of the neutral beam injection device. Description of the Drawings

[0017] Figure 1 It is a schematic structural diagram of an intelligent control system for a neutral beam injection device provided by an embodiment of the present invention; Figure 2 It is a schematic structural diagram of an intelligent control system for a neutral beam injection device provided by another embodiment of the present invention. Detailed Embodiments

[0018] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art in the technical field of the present invention without creative work based on the embodiments of the present invention belong to the scope of protection of the present invention.

[0019] The embodiment of the present invention provides an intelligent control system for a neutral beam injection device. Refer to Figure 1 As shown, it is a schematic structural diagram of an intelligent control system for a neutral beam injection device provided by an embodiment of the present invention. The intelligent control system includes a measurement and control module, a digital twin module, and a performance prediction module; The measurement and control module is used to control the coordinated operation of multiple subsystems of the neutral beam injection device during the neutral beam injection experiment and manage the experimental data generated during the neutral beam injection experiment; The digital twin module is used to construct a digital twin of the neutral beam injection device in the digital space and map the operating states of multiple subsystems of the neutral beam injection device according to the digital twin. The performance prediction module is used to predict the possibility of arcing during the neutral beam injection experiment and predict the neutral beam injection power of the neutral beam injection device under the current experimental parameters.

[0020] Specifically, the intelligent control system mainly consists of a measurement and control module, a digital twin module, and a performance prediction module. Among them, the measurement and control module is mainly responsible for controlling the coordinated operation of multiple subsystems of the neutral beam injection device during the normal experimental operation of the neutral beam injection device. For example, during the processes of source plasma generation, extraction, acceleration, and neutralization and deflection of high-energy particle beams, it coordinately controls the matching input and collaborative operation of multiple subsystems such as the power supply system, gas supply system, and deflection system of the neutral beam injection device. In addition, the measurement and control module is also responsible for managing the experimental data generated during the experimental process of the neutral beam injection device. The digital twin module is mainly responsible for constructing a digital twin of the neutral beam injection device in the digital space. In addition, the digital twin module is also responsible for mapping the operating states of multiple subsystems of the neutral beam injection device (such as multiple subsystems including ion sources, vacuum systems, high-voltage systems, and water-cooling systems) in the digital space according to the constructed digital twin to dynamically simulate and map the operating states of each physical subsystem. The performance prediction module is mainly responsible for predicting in advance the possibility of arcing during the experimental operation of the neutral beam injection device. In addition, the performance prediction module is also responsible for predicting the neutral beam injection power of the neutral beam injection device operating under the current experimental parameter conditions.

[0021] It should be noted that the measurement and control module can be designed using distributed control technology to achieve decentralized control and centralized management of the experimental operation of the neutral beam injection device. In addition, the measurement and control module can also provide a visual human-computer interaction service for the neutral beam injection device to complete the automatic preparation and task scheduling of the experimental operation of the neutral beam injection device.

[0022] It should be noted that the digital twin module can be developed and designed based on Unity, or other existing technologies can also be used for development and design. The embodiments of the present invention do not make specific limitations.

[0023] It should be noted that the performance prediction module predicts arcing and neutral beam injection power yield, which can avoid arcing during the experimental operation caused by insufficient experience of experimental personnel or mismatched experimental parameters, and can reduce ineffective shots that do not meet requirements during the operation of the neutral beam injection device.

[0024] An intelligent control system for a neutral beam injection device provided by an embodiment of the present invention can improve the safety and reliability of controlling the neutral beam injection device by integrally controlling and managing the experimental process of the neutral beam injection device and predicting the performance of the neutral beam injection device in terms of arcing and key experimental parameters. At the same time, a digital twin is also used to map the operating states of multiple subsystems of the neutral beam injection device, which can improve the timeliness and intelligent level of controlling the neutral beam injection device, thus meeting the intelligent operation requirements of the neutral beam injection device.

[0025] See Figure 2 As shown, it is a schematic structural diagram of an intelligent control system for a neutral beam injection device provided by another embodiment of the present invention. In one alternative embodiment, the measurement and control module includes: An integrated control unit, which is used to adopt distributed control technology to control the coordinated operation of multiple subsystems of the neutral beam injection device during the neutral beam injection experiment, and is also used to centrally manage the experimental parameter configuration information and experimental control process of the neutral beam injection device; A data management unit, which is used to manage the experimental data generated during the neutral beam injection experiment; wherein, the management includes data acquisition, data storage and data sharing.

[0026] Specifically, in combination with the above embodiment, the measurement and control module mainly includes an integrated control unit and a data management unit; among them, the integrated control unit mainly adopts distributed control technology and is responsible for controlling the coordinated operation of multiple subsystems of the neutral beam injection device during the normal experimental operation of the neutral beam injection device, and, the integrated control unit is also responsible for centrally managing the experimental parameter configuration information (such as clock reference, pulse timing, etc.) and experimental control process of multiple subsystems of the neutral beam injection device; the data management unit is mainly responsible for data acquisition, data storage and data sharing of the experimental data generated during the experimental process of the neutral beam injection device.

[0027] It should be noted that the integrated control unit can be developed based on the EPICS (Experimental Physics and Industrial Control System) architecture with good stability and easy maintenance, adopt a channel access mechanism, and support functions such as dynamic channel positioning, data reading and writing, access monitoring, connection monitoring and automatic reconnection, and is applicable to application scenarios such as more measurement and control positions, more types of measurement and control equipment, and different measurement and control requirements of the neutral beam injection device.

[0028] It should be noted that the integrated control unit can meet the requirements of remote visualization integrated control of the neutral beam injection device, complete the automatic preparation and task scheduling of experimental operations, and can formulate an experimental control process according to the operating characteristics of the neutral beam injection device, so as to ensure the orderly and safe development of neutral beam injection experiments; the integrated control unit can also provide a unified clock reference for each subsystem of the neutral beam injection device, so as to ensure that the experimental data of each subsystem has the same timestamp; the integrated control unit can also provide precise pulse timing for multiple subsystems such as the power supply system, diagnostic system, and gas supply system, so as to control the input time of each subsystem; the integrated control unit can also complete the process monitoring of multiple subsystems such as the vacuum system, gas supply system, and power supply and distribution system, and has on-site slow control functions such as full-graphic local / remote monitoring, setting of operating parameters and alarm thresholds, status inspection, and signal interlock between systems; the integrated control unit can also realize the control and spark management of the power supply system, and drive the power supply system to operate when special asynchronous events such as sparking occur; among them, the data interaction of experimental parameter configuration information, experimental control process, etc., which determines the safe operation of each subsystem, can adopt the Transmission Control Protocol (TCP for short), while the status information interaction with frequent interaction and many demand sides such as gun numbers, system real-time status, and waveform feedback can adopt the Channel Access (CA for short) protocol, so as to effectively ensure the reliability of the operation of each subsystem.

[0029] In one optional embodiment, the data management unit manages the experimental data generated during the neutral beam injection experiment, specifically including: For the experimental data generated during the neutral beam injection experiment, low-speed acquisition and high-speed acquisition are respectively performed according to the data acquisition requirements; The experimental data acquired at low speed and the experimental data acquired at high speed are stored in a unified data storage format, and are stored in the short term during the experimental operation and converted to long-term storage after the experiment ends; among them, the storage period of the short-term storage is less than the storage period of the long-term storage; Share the experimental data acquired at low speed and the experimental data acquired at high speed with the experimental personnel.

[0030] Specifically, in combination with the above embodiments, when the data management unit manages the experimental data generated during the experiment of the neutral beam injection device, for the experimental data generated during the experiment of the neutral beam injection device, the data management unit can perform low-speed acquisition and high-speed acquisition on different experimental data respectively according to the data acquisition requirements (for example, low-speed acquisition requirements, high-speed acquisition requirements, etc.), and adopt a unified data storage format to store the experimental data obtained by low-speed acquisition and the experimental data obtained by high-speed acquisition. The storage is divided into short-term storage and long-term storage. During the experiment operation, the experimental data obtained by low-speed acquisition and the experimental data obtained by high-speed acquisition can be stored in short-term. After the experiment is over, the short-term stored data can be converted into long-term storage. For example, after the experiment is over, the short-term stored data can be transmitted to the long-term data archive for storage to ensure that the access to historical experimental data does not affect the execution of the data archiving operation during the beam extraction experiment. It can be understood that the storage period of short-term storage is less than that of long-term storage; in addition, the data management unit can also share the experimental data obtained by low-speed acquisition and the experimental data obtained by high-speed acquisition with the experimenters. For example, provide the experimenters with tools and methods to obtain all experimental data (provide the experimenters with a graphical data access interface, an online data processing interface, an offline data access interface, etc.) to facilitate the online management and offline access of experimental data.

[0031] Exemplarily, only the low-bandwidth signals that need to be collected during each shot experiment operation phase, such as radio frequency power source, bias power supply, magnetic field power supply signals, etc., have low-speed acquisition requirements, and low-speed acquisition can be performed; the high-bandwidth signals triggered by asynchronous events such as arcing, such as acceleration power supply signals, have high-speed acquisition requirements, and high-speed acquisition can be performed.

[0032] Exemplarily, high-speed acquisition can be performed on the key signals at the moment of arcing to accurately characterize the waveform changes at the moment of arcing, so as to provide reliable data support for the analysis of the arcing cause.

[0033] It should be noted that the data management unit can be developed based on MDSplus (Model Drive System plus, a model-driven data acquisition system) to achieve low-speed and high-speed acquisition of experimental data of the neutral beam injection device, and has the function of real-time monitoring of key experimental parameters such as high-voltage and current during the experiment; the data management unit can also open different acquisition channels according to the experimental operation mode (for example, only the intake measurement channel is opened in the GAS (gas) mode, the intake, incident power, and reflected power measurement channels are opened in the RF (radio frequency) mode, and all connected measurement channels such as intake, incident power, reflected power, acceleration voltage, and acceleration current are opened in the CAL (target shooting) mode), so as to reduce the amount of data acquisition and then save data storage space; the data management unit can also use the multi-thread parallel processing method to perform segmented parallel transmission of data, so as to improve the data transmission rate and the real-time performance of data sharing; the data management unit can also achieve the alignment of the ignition moment through the hardware trigger method, which is convenient for the positioning and tracking of ignition events; the data management unit can also use the MDSplus "time slice" technology to solve the problem of real-time processing of steady-state large-scale data.

[0034] Furthermore, the data management unit also unifies the data storage format and data access format of the experimental data of the neutral beam injection device, and divides the experimental data into a tree structure for storage. Each tree contains several nodes arranged hierarchically. Users can flexibly organize the tree structure, and the segmented storage method provided by the data management unit is suitable for processing the long-pulse experimental data of the neutral beam injection device. Data from all platforms can be stored in the same format and with the same API (Application Programming Interface).

[0035] Combined with Figure 2 As shown in the figure, in one optional embodiment, the measurement and control module further includes: An interlock protection unit for interlock protection of the neutral beam injection device during the neutral beam injection experiment; wherein, the interlock protection uses optical signals as the transmission medium for fault signals.

[0036] Specifically, combined with the above embodiments, the measurement and control module includes, in addition to the integrated control unit and the data management unit, an interlock protection unit. The interlock protection unit is mainly responsible for interlock protection of the neutral beam injection device during the experiment of the neutral beam injection device to ensure the safety of the neutral beam injection device under abnormal conditions.

[0037] It should be noted that the interlock protection unit can use optical signals as the isolation transmission medium for the fault signals generated by each subsystem such as the power supply system, gas supply system, deflection system, and diagnostic system, thereby effectively improving the reliability of signal transmission.

[0038] It should be noted that the interlock protection unit can be developed based on hardware logic circuits. According to the characteristics of each different subsystem of the neutral beam injection device, the interlock objects can be divided into two categories: fast faults and slow faults. For different types of interlock signals, an FPGA (Field-Programmable Gate Array) and a PLC (Programmable Logic Controller) can be used as the main controllers respectively. The FPGA is responsible for processing the interlock signals of fast faults, and the PLC is responsible for processing the interlock signals of slow faults. Together with the underlying hardware circuit, a fast response to interlock signals can be achieved.

[0039] It should be noted that the interlock protection unit also has functions such as remote monitoring, log recording, fault latching, fault alarm, and fault tracking. For example, it can monitor the operating status of each subsystem of the neutral beam injection device (such as the power supply system, gas supply system, deflection system, etc.) in real time to perform fault alarm and interlock protection in case of a fault, facilitating fault tracking and problem troubleshooting; the interlock protection unit also has a fault masking function and can mask specific states when necessary to facilitate the preliminary debugging of the neutral beam injection device; the interlock protection unit also has an experimental operation mode recognition function and can formulate different interlock protection strategies according to different experimental operation modes, thereby providing more diverse and comprehensive protection for the safety of the neutral beam injection device under abnormal conditions and making the neutral beam injection device operate more safely and reliably.

[0040] Combined with Figure 2 As shown, in one optional embodiment, the digital twin module includes: A virtual-real data circulation unit for realizing the two-way circulation of the actual operation data of multiple subsystems of the neutral beam injection device and the virtual simulation data of the digital twin; An actual data mapping unit for mapping the actual operation data of multiple subsystems of the neutral beam injection device to the digital space; An intelligent state recognition unit for recognizing the operating status of multiple subsystems of the neutral beam injection device and adjusting the operating status of the digital twin according to the recognition result.

[0041] Specifically, in combination with the above embodiments, the digital twin module mainly includes a virtual-real data circulation unit, an actual data mapping unit and an intelligent state identification unit; wherein, the virtual-real data circulation unit is mainly responsible for realizing the bidirectional circulation of the actual operation data of multiple subsystems of the neutral beam injection device and the virtual simulation data of the digital twin; the actual data mapping unit is mainly responsible for mapping the actual operation data of multiple subsystems of the neutral beam injection device to the digital space; the intelligent state identification unit is mainly responsible for identifying the operation status of multiple subsystems of the neutral beam injection device, and adjusting the operation status of the digital twin according to the obtained identification results.

[0042] It should be noted that the virtual-reality data circulation unit plays the role of a bridge for data interaction between the physical world and the digital world. It can realize the common circulation of actual data and virtual data to open up the connection channel between reality and virtuality, and facilitate the subsequent completion of three-dimensional visualization replication, mapping and inversion of the neutral beam injection device; further, the virtual-reality data circulation unit can not only upload the real-time operation data of multiple subsystems (i.e., physical systems) of the neutral beam injection device (for example, the operation data of the ion source, the status of the high-voltage power supply, the vacuum system values ​​and the water cooling system data, etc.) to the digital twin, but also can feed back the virtual simulation results generated by the digital twin to the physical control module (for example, the measurement and control module in the embodiment of the present invention) to form a closed loop of bidirectional data flow.

[0043] Exemplarily, the core of the virtual-real data circulation unit is a time series database built based on IoTDB (Internet of Things Database).

[0044] It should be noted that the actual data mapping unit is the key hub of the digital twin module. It is responsible for accurately mapping the multi-dimensional data of the physical system into the digital space. It can map the actual operation data and status of the physical system into the digital space, which is convenient for global information grasp and real-time control. In terms of data synchronization, the actual data mapping unit can adopt multiple communication modules, including the EPICS Channel Access protocol based on experimental physics and industrial control systems, S7.Net protocol and TCP / IP (Transmission Control Protocol / Internet Protocol) protocol to complete efficient data transmission and ensure the real-time and low latency of the data. At the same time, the actual data mapping unit involves the underlying and upper-level devices, so it has a certain data fusion capability and can integrate heterogeneous data from different sources. It can improve the quality and availability of data by completing data cleaning, denoising and standardization in the performance prediction module (completed with the help of the data processing unit in the performance prediction module).

[0045] Exemplarily, the actual data mapping unit first preprocesses the original data, including data cleaning (noise suppression) and data conversion (unifying the format), and then uses real-time communication technology to dynamically bind the operation status information of the physical system with the digital twin and the information of the corresponding panel.

[0046] It should be noted that the intelligent status recognition unit is the core foundation of the digital twin module, responsible for real-time monitoring of the operation status of the physical system (for example, ion source, vacuum system, high-voltage system, water-cooling system, etc.), obtaining the actual operation data of each key physical system through deployment, and constructing a status recognition algorithm based on pattern recognition technology to analyze and process the actual operation data. This status recognition algorithm can automatically identify the operation mode of the physical system (involving the gas intake, arc voltage, radio frequency, beam extraction, and testing of the ion source), and timely adjust the simulation operation status of the digital twin according to the recognition result to ensure its high synchronization with the physical system.

[0047] Combined with Figure 2 As shown in A three-dimensional visualization unit, used for three-dimensional visualization processing of the operation status and data information of the digital twin.

[0048] Specifically, combined with the above embodiments, the digital twin module includes, in addition to the virtual-real data loop unit, the actual data mapping unit, and the intelligent status recognition unit, a three-dimensional visualization unit. The three-dimensional visualization unit is mainly responsible for three-dimensional visualization processing of the operation status and data information of the digital twin, realizing the dynamic simulation and digital operation of the digital twin.

[0049] It should be noted that the three-dimensional visualization unit can be completed based on the advanced graphics rendering engine Unity3D, constructing a high-fidelity three-dimensional virtual environment and supporting real-time dynamic rendering. The three-dimensional visualization unit provides users with an intuitive and interactive digital twin display interface. Users can intuitively observe the operation status and data information of the digital twin through this display interface, thereby further improving the human-computer interaction efficiency.

[0050] It should be noted that the three-dimensional visualization unit supports multi-level data mapping, which can not only display the macroscopic operation status of the physical system but also analyze the performance of specific components at the microscopic level, meeting the observation needs of different users.

[0051] In one optional embodiment, the digital twin has a dual operating kernel, one of which is used to serve the mapping of the operation status of multiple subsystems of the neutral beam injection device, and the other is used to serve the operation performance prediction of the neutral beam injection device.

[0052] Specifically, in combination with the above embodiments, the digital twin module constructs a digital twin of the neutral beam injection device in the digital space, which has a dual operating kernel. One operating kernel is used to provide services for mapping the operating states of multiple subsystems of the neutral beam injection device, and the other operating kernel is used to provide services for predicting the operating performance of the neutral beam injection device (for example, the operating performance of the ion source).

[0053] It should be noted that the digital twin module constructs a digital twin of the neutral beam injection device in the digital space, aiming to dynamically reflect the operating state of the physical system, feedback experimental data, and map the operating information of the physical system. It builds a coupled field integrated digital system with a dual operating kernel, and has digital scenario capabilities and artificial intelligence prediction capabilities. Among them, the subsystem based on the artificial intelligence operating kernel can provide an externalization interface for the performance prediction module to reduce experimental costs and improve experimental efficiency.

[0054] In addition, in the subsystem with artificial intelligence as the operating kernel, it not only supports the prediction of the operating performance of the ion source under single variable and multi-variable conditions, but also can understand the structural composition of the physical system through interactive operations (such as zooming, rotating, decomposing, etc.) to deeply analyze specific details.

[0055] It should be noted that the digital twin module can automatically capture and map the operating states of each key physical system through a pre-built state recognition algorithm. For virtual-real data interaction and dataset storage management, an abstract time series database is constructed through IoTDB to meet the large-capacity and real-time requirements; the digital twin module can also use a self-designed data mapping algorithm to map the real-time operating data to the front-end three-dimensional control panel to provide an intuitive interaction interface for experimental personnel and at the same time provide a dynamic three-dimensional visual effect for the ion source.

[0056] Combined Figure 2 As shown in the following, in one optional embodiment, the performance prediction module includes: A fire prediction unit, which is used to predict the possibility of a fire occurring during the neutral beam injection experiment according to a trained fire prediction model; A power prediction unit, which is used to predict the neutral beam injection power of the neutral beam injection device under the current experimental parameters according to a trained power prediction model.

[0057] Specifically, in combination with the above embodiments, the performance prediction module mainly includes a spark prediction unit and a power prediction unit. Among them, the spark prediction unit is mainly responsible for predicting in advance the possibility of sparking during the experimental operation of the neutral beam injection device according to the trained spark prediction model. The power prediction unit is mainly responsible for predicting the neutral beam injection power when the neutral beam injection device operates under the current experimental parameter conditions according to the trained power prediction model.

[0058] It should be noted that for the classification problem of spark prediction of the neutral beam injection device, combined with the machine learning method, a spark prediction model based on random forest is constructed and the model is trained to accurately evaluate the possibility of causing sparks in a specific experimental environment.

[0059] Exemplarily, the specific implementation process of accurately evaluating the possibility of causing sparks in a specific experimental environment is as follows: First, obtain the required historical experimental data (the historical experimental data to be obtained includes: radio frequency power supply set power, incident power, reflected power, magnetic field current, air pressure, bias voltage, bias plate current, extraction voltage, acceleration voltage, extraction current, co-extracted electron current, H-ion current, and pulse width). Then, process the obtained historical experimental data, that is, classify the type of each historical experimental data into sparking and non-sparking. After that, instead of directly inputting the processed historical experimental data into the spark prediction model, the ion source working set parameters (including: radio frequency power supply power, magnetic field current, air pressure, bias voltage, extraction voltage, and acceleration voltage) are input into the trained spark prediction model, and the model will determine whether the ion source under this set of parameters will have a sparking phenomenon.

[0060] It should be noted that for the non-linear regression problem of power prediction of the neutral beam injection device, combined with the deep learning method, power prediction models based on a multi-layer perceptron neural network with backpropagation, a deep neural network based on the attention mechanism, and a long short-term memory network based on the attention mechanism are respectively constructed. The generalization effects of different models are compared through the validation set and the test set, and then the deep neural network model based on the attention mechanism with the best prediction performance is selected as the final power prediction model to accurately estimate the neutral beam injection power yield under normal beam extraction conditions.

[0061] Exemplarily, the specific implementation process for accurately estimating the neutral beam injection power yield under normal beam extraction is as follows: First, obtain the required historical experimental data (the historical experimental data to be obtained includes: the set power of the radio frequency power supply, incident power, reflected power, magnetic field current, gas pressure, bias voltage, bias plate current, extraction voltage, acceleration voltage, extraction current, co-extracted electron current, H- ion current, and pulse width); then, take the mean value of the obtained historical experimental data after the extraction voltage is input, and train the power prediction model with these mean value data; after evaluating the model effect, by inputting the ion source working set parameters (including: radio frequency power supply power, magnetic field current, gas pressure, bias voltage, extraction voltage, and acceleration voltage) into the trained power prediction model, the model will predict the neutral beam injection power yield under these set parameters.

[0062] Combined with Figure 2 As shown, in one of the alternative embodiments, the performance prediction module further includes: A data processing unit, configured to construct a spark prediction data set and a power prediction data set, so as to train a preset spark prediction model according to the spark prediction data set, and train a preset power prediction model according to the power prediction data set.

[0063] Specifically, combined with the above embodiments, the performance prediction module, in addition to including a spark prediction unit and a power prediction unit, further includes a data processing unit, which is mainly responsible for constructing a spark prediction data set and a power prediction data set, so as to train a preset spark prediction model according to the constructed spark prediction data set, and train a preset power prediction model according to the constructed power prediction data set.

[0064] It should be noted that the spark prediction data set is mainly used to train, verify, and evaluate whether the current parameter combination may cause a spark phenomenon, and the power prediction data set is mainly used to train, verify, and evaluate the injection power that may be generated under the state where the current parameters are highly unlikely to cause a spark; it can be understood that in order to reduce the generalization error of the model, the prediction data sets (i.e., the spark prediction data set and the power prediction data set) can be divided into a training set, a validation set, and a test set. The training set is used in the model training stage, the validation set is used to evaluate the model effect and adjust the hyperparameters of the model to make the model work best on the validation set, and the test set is used to obtain some final evaluation metrics.

[0065] It should be noted that during the training process of the spark prediction model, the model performance of the spark prediction model can be evaluated through a confusion matrix, accuracy, precision, recall rate, F1 score (used to comprehensively consider the recognition ability and coverage ability of the model, which is the harmonic mean of the precision rate and the recall rate), and ROC curve (plotting the relationship between the true positive rate and the false positive rate at different thresholds to discriminate the classification performance of the model).

[0066] In one optional embodiment, the data processing unit constructs a spark prediction data set and a power prediction data set, specifically including: According to the operating principle of the neutral beam injection device, key parameters are selected as data set features; wherein, the key parameters include parameters of a radio frequency power source, an extraction power source, an acceleration power source, a bias power source, a magnetic field power source, and an intake air volume; Analyze the historical experimental data corresponding to the data set features to obtain data set feature values; Filter the data set feature values according to a preset filtering rule to construct a spark prediction data set and a power prediction data set.

[0067] Specifically, in combination with the above embodiment, when the data processing unit constructs a spark prediction data set and a power prediction data set, it can first select key parameters as data set features according to the operating principle of the neutral beam injection device. The key parameters include parameters of a radio frequency power source, an extraction power source, an acceleration power source, a bias power source, a magnetic field power source, and an intake air volume; then analyze the historical experimental data corresponding to the data set features (i.e., key parameters) to obtain the data set feature values corresponding to the data set features (i.e., key parameter feature values); finally, filter the obtained data set feature values according to the preset filtering rule to obtain a spark prediction data set and a power prediction data set respectively.

[0068] Exemplarily, the data processing unit selects key parameters such as a radio frequency power source, an extraction power source, an acceleration power source, a bias power source, a magnetic field power source, and an intake air volume as data set features according to the operating principle of the neutral beam injection device, and uses a segmented calculation method to analyze the historical experimental data corresponding to the key parameters stored in the MDSplus database to obtain key parameter feature values, which can ensure the accuracy and integrity of the structured data set of the experimental data of the neutral beam injection device, the consistency of the data format, the traceability of the data, and the compliance and security of the data; then, using network communication technology, send the calculated key parameter feature values to the data automatic recording terminal through the User Datagram Protocol (UDP) to realize the automatic recording of experimental data; finally, for the automatically recorded experimental data, construct a spark prediction data set and a power prediction data set by setting filtering rules such as beam pulse width length.

[0069] It should be noted that by constructing a spark prediction data set and a power prediction data set, the data processing unit can provide a reliable data source for subsequent spark prediction and power prediction.

[0070] It should be noted that the performance prediction module, based on artificial intelligence technologies such as machine learning and deep learning, deeply analyzes and mines the historical experimental data accumulated during the experimental operation of the neutral beam injection device, extracts the hidden laws and characteristics of the historical experimental data, and can predict the effectiveness of parameter settings before the start of the neutral beam injection experiment, reducing the probability of damage to experimental equipment caused by arcing events due to mismatched experimental parameters and extending the service life of the equipment. At the same time, it can quickly and accurately determine arcing events in the neutral beam injection device, providing a new perspective and in-depth physical explanation for understanding the working principle of arcing in the neutral beam injection device; for neutral beam injection experiments under the condition of matched experimental parameters, a performance prediction model is established through deep learning methods, which can predict the neutral beam injection power yield under the current experimental parameter conditions, reduce invalid shots that do not meet the requirements during the operation of the neutral beam injection experiment, and provide guidance for parameter adjustment, assisting experimental operators to more precisely adjust the experimental plan and resource allocation, avoiding unnecessary resource waste, and accelerating the research and development process of the neutral beam injection device.

[0071] In summary, the intelligent control system of a neutral beam injection device provided by the embodiments of the present invention adopts distributed control technology, combines the traditional control system with digital twin technology and artificial intelligence technology, realizes decentralized control and centralized management, data management, and interlock protection of the experimental operation of the neutral beam injection device, and on the basis of meeting the basic control requirements, data management requirements, and interlock protection requirements for the operation of the neutral beam injection device, realizes three-dimensional visual digital expression of the experimental operation of the neutral beam injection device and advance prediction of the performance of key parameters, meeting the high requirements of the operation of the neutral beam injection device for the safety, reliability, timeliness, and intelligent level of the control system.

[0072] The above are only the preferred embodiments of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the technical principle of the present invention, several improvements and deformations can still be made, and these improvements and deformations should also be regarded as the protection scope of the present invention.

Claims

1. An intelligent control system for a neutral beam injection device, characterized in that, It includes a measurement and control module, a digital twin module, and a performance prediction module; The measurement and control module is used to control the coordinated operation of multiple subsystems of the neutral beam injection device during the neutral beam injection experiment and manage the experimental data generated during the neutral beam injection experiment; The digital twin module is used to construct a digital twin of the neutral beam injection device in the digital space and map the operating states of multiple subsystems of the neutral beam injection device according to the digital twin; The performance prediction module is used to predict the possibility of arcing during the neutral beam injection experiment and predict the neutral beam injection power of the neutral beam injection device under the current experimental parameters.

2. The intelligent control system of the neutral beam injection device according to claim 1, characterized in that The measurement and control module includes: An integrated control unit, which is used to control the coordinated operation of multiple subsystems of the neutral beam injection device during the neutral beam injection experiment by using distributed control technology, and is also used to centrally manage the experimental parameter configuration information and experimental control process of the neutral beam injection device; A data management unit, which is used to manage the experimental data generated during the neutral beam injection experiment; wherein, the management includes data acquisition, data storage, and data sharing.

3. The intelligent control system of the neutral beam injection device according to claim 2, wherein The data management unit manages the experimental data generated during the neutral beam injection experiment, specifically including: For the experimental data generated during the neutral beam injection experiment, low-speed acquisition and high-speed acquisition are respectively performed according to the data acquisition requirements; The experimental data acquired at low speed and the experimental data acquired at high speed are stored in a unified data storage format, and are stored in the short term during the experiment operation and converted to long-term storage after the experiment ends; wherein, the storage period of the short-term storage is less than the storage period of the long-term storage; Share the experimental data acquired at low speed and the experimental data acquired at high speed with the experimenters.

4. The intelligent control system of the neutral beam injection device according to claim 2, wherein The measurement and control module further includes: An interlock protection unit, which is used to perform interlock protection on the neutral beam injection device during the neutral beam injection experiment; wherein, the interlock protection uses optical signals as the transmission medium for fault signals.

5. The intelligent control system of the neutral beam injection device according to claim 1, characterized in that, The digital twin module includes: A virtual-real data circulation unit, which is used to realize the two-way circulation of the actual operation data of multiple subsystems of the neutral beam injection device and the virtual simulation data of the digital twin; An actual data mapping unit, which is used to map the actual operation data of multiple subsystems of the neutral beam injection device to the digital space; An intelligent state recognition unit, which is used to recognize the operating states of multiple subsystems of the neutral beam injection device and adjust the operating state of the digital twin according to the recognition results.

6. The intelligent control system of the neutral beam injection device according to claim 5, characterized in that, The digital twin module further includes: A three-dimensional visualization unit, which is used to perform three-dimensional visualization processing on the operating state and data information of the digital twin.

7. The intelligent control system of the neutral beam injection device according to claim 1, characterized in that, The digital twin has a dual operating kernel, one of which is used to serve the mapping of the operating states of multiple subsystems of the neutral beam injection device, and the other is used to serve the operating performance prediction of the neutral beam injection device.

8. The intelligent control system of the neutral beam injection device according to claim 1, characterized in that, The performance prediction module includes: An arcing prediction unit, which is used to predict the possibility of arcing during the neutral beam injection experiment according to the trained arcing prediction model; A power prediction unit, configured to predict the neutral beam injection power of the neutral beam injection device under current experimental parameters according to a trained power prediction model.

9. The intelligent control system of the neutral beam injection device according to claim 8, characterized in that, The performance prediction module further includes: A data processing unit, configured to construct a spark prediction data set and a power prediction data set, so as to train a preset spark prediction model according to the spark prediction data set, and train a preset power prediction model according to the power prediction data set.

10. The intelligent control system of the neutral beam injection device according to claim 9, characterized in that, The data processing unit constructs a spark prediction data set and a power prediction data set, specifically including: According to the operating principle of the neutral beam injection device, key parameters are selected as data set features; wherein, the key parameters include parameters of a radio frequency power source, an extraction power source, an acceleration power source, a bias power source, a magnetic field power source, and an intake air volume. Analyze the historical experimental data corresponding to the data set features to obtain data set feature values. Filter the data set feature values according to a preset filtering rule to construct a spark prediction data set and a power prediction data set.

Images