Plasma control method, plasma control device, and program
Patent Information
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- ALPHADX CO LTD
- Filing Date
- 2026-01-19
- Publication Date
- 2026-06-16
AI Technical Summary
【0018】 本発明は、核融合装置において生成されるプラズマの挙動を、複数の制御アクチュエータの操作パラメータ間の均衡と相互関係を維持しつつ制御するため、魔方陣の数学的特性に基づいて当該パラメータ群の割当て規則(マッピング規則)を更新する協調制御戦略決定ロジックを中核とする。さらに、プラズマ状態量·外乱·境界条件の因果関係を因果ループダイアグラム(CLD)として表現し、強化ループおよび抑制ループの寄与度に基づくループ制約を上位制約として導入する構成とすることで、探索的不安定性を回避しながら多目的な運転目標を両立する。さらに、本発明では、乱流が有する局所的特性および非局所的特性を、因果ループダイアグラムの構造およびループ寄与度の双方に反映させることで、短時間で顕在化する不安定性の予兆に対しても、協調制御戦略決定ロジックによる迅速な対応を可能とする。
Smart Images

Figure 0007874378000005 
Figure 0007874378000006 
Figure 0007874378000007
Abstract
Claims
1. A method for controlling the behavior of plasma generated in a nuclear fusion device, We acquired multiple measurement values indicating the plasma state. Taking the acquired measured values and target states or constraints related to plasma stability as input, a cooperative control strategy decision logic is executed that represents a group of operating parameters of multiple control actuators, including a magnetic field coil, a heating device, and a fuel supply device, as a matrix structure based on the mathematical properties of a magic square, updates mapping rules that represent the balance, distribution, or interrelationships of the group of operating parameters so as to satisfy the structural constraints of the matrix structure, and outputs updated information on the arrangement or mapping of the group of operating parameters based on the cooperative control strategy decision logic. According to the output update information, the operation command value for each of the multiple control actuators is calculated. The operation command value is applied to the plurality of control actuators to control the behavior of the plasma. The aforementioned magic square has a structure in which the sum of the elements in each row, column, and diagonal is constant, and the plasma control method is characterized by using this structure to maintain balance between multiple control actuators.
2. The plasma control method according to claim 1, characterized in that the cooperative control strategy decision logic uses any of constraint optimization, model predictive control, rule-based inference, imitation learning, or evolutionary algorithms.
3. A process for generating or updating a causal loop diagram (CLD) representing the interaction of plasma state variables, multiple control actuators, and disturbance / boundary conditions, The plasma control method according to claim 1, further comprising the step of quantifying the loop contributions of the strengthening loop and the suppression loop in the causal loop diagram.
4. The plasma control method according to claim 3, characterized in that, in the step of generating or updating the causal loop diagram, local heat transport characteristics and non-local thermal coupling characteristics due to turbulence are expressed as causal relationships, and in the step of quantifying the loop contribution, the contribution of the causal relationship corresponding to the local heat transport characteristics and the non-local thermal coupling characteristics is estimated and reflected in the control.
5. The plasma control method according to claim 3, characterized in that the loop constraints obtained from the causal loop diagram are given as higher-level constraints, and the operation command values are calculated under the constraint that the sum of the rows, columns, and diagonals of the magic square is constant.
6. The plasma control method according to claim 3, characterized in that, upon detection of an instability event, the contributions of the strengthening loop and suppression loop in the causal loop diagram are reestimated by analyzing log data, the causal loop diagram is updated, and the mapping rules and operation command values of the magic square are recalculated based on the update results.
7. The plasma control method according to claim 1, characterized in that the cooperative control strategy decision logic does not use maximizing a scalar reward signal based on trial and error with the environment as its primary learning or optimization principle.
8. The plasma control method according to claim 1, characterized in that the magnetic field coils included in the plurality of control actuators include a toroidal coil, a poloidal coil, or a vertical magnetic field coil.
9. A program for causing a computer to perform the method according to any one of claims 1 to 8.
10. The program according to claim 9, characterized in that it uses a data access library for a data model that conforms to the interface data structure (IDS) defined by the integrated modeling and analysis suite to acquire an input IDS including the measured values and write an output IDS storing the operation command values.
11. A plasma control device for controlling the behavior of plasma generated in a nuclear fusion device, A measurement value acquisition unit that acquires multiple measurement values indicating the plasma state, A cooperative control strategy decision logic unit takes the measured values acquired by the measurement value acquisition unit and the target state or constraints related to plasma stability as input, represents a group of operation parameters of multiple control actuators, including a magnetic field coil, a heating device, and a fuel supply device, as a matrix structure based on the mathematical properties of a magic square, updates mapping rules that express the balance, distribution, or interrelationships of the group of operation parameters so as to satisfy the structural constraints of the matrix structure, and outputs updated information on the arrangement or correspondence of the group of operation parameters. A command value calculation unit calculates an operation command value for each of the multiple control actuators according to the update information, An actuator output unit that applies the operation command value to the plurality of control actuators, Equipped with, A plasma control device characterized in that the magic square has a structure in which the sum of the elements in the rows, columns, and diagonals is constant, and this structure is used to maintain balance between a plurality of control actuators.
12. The plasma control device according to claim 11, characterized in that the cooperative control strategy decision logic unit updates the mapping rules using at least one of constraint optimization, model predictive control, rule-based inference, imitation learning, or evolutionary algorithms.
13. A CLD generation / update unit that generates or updates a causal loop diagram (CLD) of the interaction between plasma state variables, multiple control actuators, and disturbance / boundary conditions, A loop contribution estimation unit that quantifies the contributions of strengthening loops and suppression loops within the causal loop diagram. The plasma control device according to claim 11, further comprising:
14. The CLD generation and renewal unit expresses the local heat transport characteristics and non-local thermal coupling characteristics due to turbulence as a causal relationship. The plasma control device according to claim 13, characterized in that the loop contribution estimation unit estimates the contribution of causal relationships corresponding to the local heat transport characteristics and the non-local thermal coupling characteristics and reflects this in the control.
15. The plasma control device according to claim 13, characterized in that the command value calculation unit calculates an operation command value under the constraint that the sum of the rows, columns, and diagonals of the magic square is constant, while providing the loop constraint obtained from the causal loop diagram as a higher-level constraint.
16. It further includes an instability event detection unit that detects instability events, The loop contribution estimation unit, upon detection of an instability event by the instability event detection unit, re-estimates the contributions of the reinforcing loop and suppression loop in the causal loop diagram. The CLD generation and update unit updates the causal loop diagram based on the contributions of the reinforcing loops and suppressing loops in the causal loop diagram re-estimated by the loop contribution estimation unit. The aforementioned cooperative control strategy decision logic unit updates the mapping rules again based on the update results. The plasma control device according to claim 13, characterized in that the command value calculation unit recalculates the operation command value.
17. The plasma control device according to claim 11, characterized in that the cooperative control strategy decision logic unit does not use maximizing a scalar reward signal based on trial and error with the environment as its primary learning or optimization principle.
18. The plasma control device according to claim 11, characterized in that the magnetic field coils included in the plurality of control actuators include toroidal coils, poloidal coils, or vertical magnetic field coils.
19. The plasma control device according to claim 11, characterized in that the measurement value acquisition unit, the cooperative control strategy determination logic unit, the command value calculation unit, and the actuator output unit input and output the measurement value and the operation command value using a data model defined by an interface data structure (IDS) defined by an integrated modeling and analysis suite.
20. The plasma control device according to claim 19, characterized in that the measurement value acquisition unit acquires an input IDS storing the measurement value using a data access library conforming to the data model, and the command value calculation unit writes an output IDS storing the operation command value.
21. The plasma control device according to claim 19, characterized in that the command value calculation unit reads the input IDS in object format, executes a predetermined control algorithm, and writes the resulting operation command value to the output IDS in object format.